Datasets:

---

bigcode

/

jupyter-parsed

like 4

Follow

BigCode 1.36k

[ "MIT" ]

[ "MIT" ]

[ "MIT" ]

[ [ [ "# Radioactive Decay Interactives\n\n## Interactive Figure 1: Model of Radioactive Decay\n\nThis first figure takes a population of 900 atoms and models their radioactive decay.", "_____no_output_____" ], [ "This first interactive is designed to allow you to explore what happens during radioactive decay of some isotope. An *isotope* refers to a particular version of an element. For example, the non-radioactive Carbon-12 isotope has 6 protons and 6 neutrons in its nucleus but the radioactive Carbon-14 isotope has 6 protons and 8 neutrons in its nucleus. Different isotopes of the same element behave identically in terms of chemistry and bonding to other atoms, but their nuclear properties can differ.\n\nDuring radioactive decay, a *parent isotope* is said to decay into a *daughter isotope*. So, for example, the parent isotope of Carbon-14 decays into Nitrogen-14. There is no way to predict when any one nucleus of a *parent isotope* will decay into a *daughter isotope*. That said, if you look at a large number of nuclei of a parent isotope, they exhibit a very simple property:\n\n> **The same *fraction* of a radioactive parent isotope will decay over the same amount of time.**\n \nThe time it takes for one-half of a population of parent isotopes to decay (on average) into daughter isotopes is called the *half-life* of that isotope. Different parent isotopes can have very different half-lifes. **NOTE:** This interactive shows a simulation of the decay of only 900 atoms. The radioactive decay is still modelled as occurring randomly for any one atom, so the simulation will show slightly different results on different runs!!", "_____no_output_____" ], [ "Some questions to consider \n1. After 1 half-life, about 50% of a parent isotope should have decayed and become daughter isotope. Use the interactive graphic below and adjust the elapsed time to figure out how long one half-life is for each isotope. Explain your approach.\n2. How much of a parent isotope is left after 2 half-lifes? 3 half-lifes? 4 half-lifes? 5 half-lifes? Explain how you figured this out.\n3. You should have found about 25% of the original amount of parent isotope is left after 2 half-lifes. This may seem surprising since in the first half-life 50% of the original amount of parent isotope decayed. Explain why 'less' of it decayed during the second half-life. **HINT** Consider the very simple property of radioactive decay we highlighted above.", "_____no_output_____" ] ], [ [ "from IPython.display import display\nimport numpy as np\nimport bqplot as bq\nimport ipywidgets as widgets\nimport random as random\nimport pandas as pd\nimport number_formatting as nf\nfrom math import ceil, floor, log10", "_____no_output_____" ], [ "## Originally developed June 2018 by Samuel Holen\n##\n## Edits by Juan Cabanela October 2018 to allow changes in the GUI.\n## - Made the display of the precise number/faction of parent/daught atoms instructor\n## configurable.\n## - Fixed a problem with the display of data, forced data to only be displayed for first\n## 10 half-lifes regardless of actual generated decay times (this allows hard-coding of)\n## num_ticks and tick values.\n\n## Pre-construct model of radioactive decay of a population\n## of parent and daughter atoms.\n## \n\n# GUI Configuration Parameters\nshow_counts = True\nmax_half_lifes = 10 # maximum half-lives to graph\ntime_ticks = 6 # number of ticks for the time axis\n# half-lives to place ticks on horizontal axis\nhalf_life_ticks = np.linspace(0, max_half_lifes, time_ticks)\n\n# Constants Related to decay of the parent species to the daughter species\nN_parent = 900 # initial number of parent atoms (should be a perfect square)\nN_daughter = 0 # initial number of daughter atoms\ntau = 1 # placeholder for the half-life of the parent species \nh = 0.025 # time step (in half lives)\nmu = np.log(2.) / tau # constant for decay time distribution \nPlot_all_times = True # Plot all times (otherwise, selects only before time slider value)\n\n# Initialize tracking of number of atoms\nParent_counts = [] # list of number of parent atoms \nDauther_counts = [] # list of number of daughter atoms\n\n# Generate a uniform random distribution of N_parent numbers from 0 to 1\nz = np.random.rand(N_parent)\n\n# Function to convert uniform distribution of random numbers to\n# a distribution weighted to model radiactive decay. The times are in number of \n# half-lives. \n#\n# The unsorted data representing the number of half-lifes until the individual\n# decay of each atom.\ndecay_times = -np.log(1 - z) / mu\ndecay_times_sorted = np.sort( decay_times )\n\n# Genereate array of numbers of atoms left\n# Adjusted so that each count contains 0 and N_parent\nParent_counts = np.arange(N_parent,-1, -1, dtype='int') # Number of parent atoms\nDaughter_counts = np.ones_like(Parent_counts)\nDaughter_counts = N_parent - Parent_counts # Number of daughter atoms\n\n#\n# Construct Pandas data frames\n#\n\n# Time column adjusted to include t=0\ndecay_data = pd.DataFrame()\ndecay_data['time'] = np.concatenate((np.zeros(1), decay_times_sorted))\ndecay_data['Parent'] = Parent_counts\ndecay_data['Daughter'] = Daughter_counts\n\n# Data array for species\nspecies = pd.DataFrame()\nspecies['parent_long'] = ['Generic', 'Carbon', 'Thallium','Uranium','Rubidium']\nspecies['daughter_long'] = ['Generic', 'Nitrogen','Lead','Thorium','Strontium']\nspecies['parent_short'] = ['Parent', 'C-14','Tl-208','U-235','Rb-87']\nspecies['daughter_short'] = ['Daughter', 'N-14','Pb-208','Th-231','Sr-87']\nspecies['half-lives'] = [tau, 5730, 3.053 * 60, 703.8, 48.8]\nspecies['step-size'] = [h, 125, 0.05 * 60, 15, 1]\nspecies['timeunits'] = ['half-lives', 'years', 'seconds', 'million years', 'billion years']", "_____no_output_____" ], [ "##\n## Define functions to respond to controls on population plots.\n##\n\ndef UpdateSpecies(change=None):\n ##\n ## Deal with possible changes of species\n ##\n \n # Generate a new uniform random distribution of N_parent numbers from 0 to 1 and\n # then convert uniform distribution to one weighted to model radioactive decay.\n z = np.random.rand(N_parent)\n decay_times = -np.log(1 - z) / mu\n decay_times_sorted = np.sort( decay_times )\n decay_data['time'] = np.concatenate((np.zeros(1), decay_times_sorted))\n \n # Reset the time to zero\n Time_slide.value = 0\n \n # Adjust half-life data and limits on plots appropriately \n new_index = species.loc[species.parent_long == pick_Species.value].index[0]\n \n # Set the half-life based the selected species\n hf = species['half-lives'][new_index]\n \n # Set the limit on the time sider to be 10 half-lifes\n Time_slide.max = max_half_lifes*hf\n \n # Set the time step on slider\n Time_slide.step = species['step-size'][new_index]\n\n # Updates the time slider label/units\n unit_label.value = species['timeunits'][new_index]\n \n # Updates the time axes on the plot\n x_time.max = Time_slide.max\n ax_x_time.scale = x_time\n ax_x_time.label = unit_label.value\n \n # Update tick values\n ax_x_time.tick_values = list(half_life_ticks*hf)\n\n # Update the legend\n parent_label_new = species['parent_short'][new_index]\n daughter_label_new = species['daughter_short'][new_index]\n line_parent.labels = [parent_label_new]\n line_daughter.labels = [daughter_label_new]\n \n # Update the species in the box that shows how many are present \n parent_label.value = parent_label_new + ' produced'\n daughter_label.value = daughter_label_new + ' remaining'\n\n # Now call the function for updating the plot in the event of a time change\n UpdateTimes()\n\n \ndef UpdateFraction(change=None):\n ##\n ## Deal with whether fraction view or number view selected to set scales for plot,\n ## y values for plot, and label values\n ##\n if frac_or_num.value == False:\n # Number count mode enabled\n\n # Update axes and scales\n fig_counts.axes = [ax_x_time, ax_y_number]\n line_parent.scales={'x': x_time, 'y': y_number}\n line_daughter.scales={'x': x_time, 'y': y_number}\n pts_parent.scales={'x': x_time, 'y': y_number}\n pts_daughter.scales={'x': x_time, 'y': y_number}\n line_time.scales={'x': x_time, 'y': y_number}\n\n # Update 'time line' limits\n line_time.y = [0, N_parent] \n else:\n # Fraction mode enabled\n\n # Updated axes and scales\n fig_counts.axes = [ax_x_time, ax_y_fraction]\n line_parent.scales={'x': x_time, 'y': y_fraction}\n line_daughter.scales={'x': x_time, 'y': y_fraction}\n pts_parent.scales={'x': x_time, 'y': y_fraction}\n pts_daughter.scales={'x': x_time, 'y': y_fraction}\n line_time.scales={'x': x_time, 'y': y_fraction}\n\n # Update 'time line' limits\n line_time.y = [0, 1]\n\n UpdateTimes()\n\n \ndef UpdateTimes(change=None):\n ##\n ## Deal with changes in the time slider\n ##\n\n # Recall half-life of this species\n species_idx = species.loc[species.parent_long == pick_Species.value].index[0]\n hf = species['half-lives'][species_idx]\n\n # Update time label with 3 significant figures\n Time_label.value = str(nf.SigFig(Time_slide.value, 3))\n \n # Get the array of times\n time_arr = hf * decay_data['time']\n \n # Set the times to be x values\n line_daughter.x = time_arr\n line_parent.x = line_daughter.x\n pts_daughter.x = line_daughter.x\n pts_parent.x = line_daughter.x\n line_time.x = [Time_slide.value, Time_slide.value]\n \n # Changes the color of population to reflect the decays\n for i in range(N_parent):\n if Time_slide.value >= decay_times[i]*hf:\n Colors[i] = 'blue'\n else:\n Colors[i] = 'red'\n population_scat.colors = Colors\n \n # Identify where we are in the data set\n i = 0\n while i < N_parent + 1 and hf*decay_data['time'][i] < Time_slide.value: \n i += 1\n if i > 0:\n i -= 1\n \n # Update selection of the parent and daughter data to be plotted\n if Plot_all_times:\n daughter_decay = decay_data['Daughter']\n parent_decay = decay_data['Parent']\n else:\n daughter_decay = decay_data['Daughter'][0:i+1]\n parent_decay = decay_data['Parent'][0:i+1] \n \n line_parent.y = parent_decay\n line_daughter.y = daughter_decay\n \n ##\n ## Deal with whether fraction view or number view selected to set scales for plot,\n ## y values for plot, and label values\n ##\n if frac_or_num.value == False:\n # Number count mode enabled\n\n # Update number of parent and daughter in labels\n parent_present.value = str(decay_data['Parent'][i])\n daughter_present.value = str(decay_data['Daughter'][i])\n else:\n # Fraction mode enabled\n \n # Update the x and y arrays for the parent and daughter lines\n line_parent.y = (1/N_parent)*parent_decay\n line_daughter.y = (1/N_parent)*daughter_decay\n \n # Update number of parent and daughter in labels\n parent_present.value = '{:.3f}'.format((1/N_parent)*decay_data['Parent'][i])\n daughter_present.value = '{:.3f}'.format((1/N_parent)*decay_data['Daughter'][i])\n \n # Update parent and daughter lines\n pts_parent.y = line_parent.y\n pts_daughter.y = line_daughter.y\n \n # Update the tooltip labels and formats (Bug? Doesn't appear to be working)\n #pts_parent.tooltip.labels[1] = parent_label_new\n #pts_daughter.tooltip.labels[1] = daughter_label_new\n #if frac_or_num.value == False:\n # pts_parent.tooltip.formats[1] = '3.0f'\n # pts_daughter.tooltip.formats[1] = '3.0f'\n #else:\n # pts_parent.tooltip.formats[1] = '0.3f'\n # pts_daughter.tooltip.formats[1] = '0.3f'\n \n", "_____no_output_____" ], [ "##\n## Set up counts versus time plot\n##\n\n# Set up Species to be Generic\ninit_species_ind = 0 \n\n# Set up initial time\ninit_time = 0\ninit_time_idx = 0\n\n# Set up axes\nx_time = bq.LinearScale(min = 0, max=max_half_lifes)\ny_number = bq.LinearScale(min = 0, max=N_parent)\ny_fraction = bq.LinearScale(min = 0, max=1)\n\n# Labels and scales for Axes\nax_x_time = bq.Axis(label=species['timeunits'][init_species_ind], scale=x_time, num_ticks = time_ticks,\n tick_values = half_life_ticks )\nax_y_number = bq.Axis(label='Number of atoms', scale=y_number, orientation='vertical')\nax_y_fraction = bq.Axis(label='Fraction of atoms', scale=y_fraction, orientation='vertical')\n\n# Define tooltip (Bug: doesn't allow relabeling Tooltips, also needs to apply to Scatter, not Lines)\n#def_tt_parent = bq.Tooltip(fields=['x', 'y'], formats=['.2f', '3.0f'], labels=['time', species['parent_short'][init_species_ind]])\n#def_tt_daughter = bq.Tooltip(fields=['x', 'y'], formats=['.2f', '3.0f'], labels=['time', species['daughter_short'][init_species_ind]])\ndef_tt_parent = bq.Tooltip(fields=['x', 'y'], formats=['.2f', '.3f'], labels=['time', 'amount of parent isotope'])\ndef_tt_daughter = bq.Tooltip(fields=['x', 'y'], formats=['.2f', '.3f'], labels=['time', 'amount of daughter isotope'])\n\n# Define the Lines and Scatter plots\n# NOTE: Scatter only necessary to allow tooltips to function.\ninit_x = decay_data['time']*species['half-lives'][init_species_ind]\nif Plot_all_times:\n init_parent = decay_data['Parent']\n init_daughter = decay_data['Daughter']\nelse:\n init_parent = decay_data['Parent'][0:init_time_idx]\n init_daughter = decay_data['Daughter'][0:init_time_idx]\n \npts_parent = bq.Scatter(x=init_x, y=init_parent,\n scales={'x': x_time, 'y': y_number}, marker='circle', default_size=2,\n display_legend=False, colors=['red'], labels=[species['parent_short'][init_species_ind]], \n tooltip=def_tt_parent)\npts_daughter = bq.Scatter(x=init_x, y=init_daughter,\n scales={'x': x_time, 'y': y_number}, marker='circle', default_size=2,\n display_legend=False, colors=['blue'], labels=[species['daughter_short'][init_species_ind]],\n tooltip=def_tt_daughter)\nline_parent = bq.Lines(x=init_x, y=init_parent, \n scales={'x': x_time, 'y': y_number}, display_legend=True, colors=['red'], \n labels=[species['parent_short'][init_species_ind]], )\nline_daughter = bq.Lines(x=init_x, y=init_daughter, \n scales={'x': x_time, 'y': y_number}, display_legend=True, colors=['blue'], \n labels=[species['daughter_short'][init_species_ind]] )\n\n# Set up a vertical line on this plot to indicate the current time\ntimes_x = [init_time, init_time]\ntimes_y = [0, N_parent]\nline_time = bq.Lines(x=times_x, y=times_y,\n scales={'x': x_time, 'y': y_number}, \n colors=['greenyellow'])\n\n# Creates figure for plot\nfig_counts = bq.Figure(axes=[ax_x_time, ax_y_number], marks=[line_parent, line_daughter, line_time, pts_parent, pts_daughter], \n legend_location='right', legend_style={'fill': 'white'}, \n title='Counts versus Time', background_style={'fill': 'black'}, \n layout={'width': '500px', 'min_height': '400px'},\n animation=1000)", "_____no_output_____" ], [ "# Slider widget to control the amount of time that has passed\n# Set for generic half-life situation initially (0 to 10 half-lifes)\nTime_slide = widgets.FloatSlider(\n value=0.,\n description='Time',\n min=0.,\n max=max_half_lifes,\n step=h,\n disabled=False,\n continuous_update=False,\n orientation='horizontal',\n readout=False,\n readout_format='.1f',\n layout=widgets.Layout(overflow_x='visible',\n overflow_y='visible',\n width='400px',\n max_width='500px',\n min_width='250px')\n)\n\n# Widget to display the number of parent atoms present\nparent_present = widgets.Text(\n value = str(N_parent),\n style = {'description_width': 'initial'},\n #description = species['parent_short'][0]+' remaining',\n disabled = True,\n layout=widgets.Layout(overflow_x='visible',\n overflow_y='visible',\n width='175px',\n max_width='300px',\n min_width='125px')\n)\n\n# Widget to display the number of daughter atoms present\ndaughter_present = widgets.Text(\n value = str(0),\n style = {'description_width': 'initial'},\n #description = species['daughter_short'][0]+' produced',\n disabled = True,\n layout=widgets.Layout(overflow_x='visible',\n overflow_y='visible',\n width='175px',\n max_width='300px',\n min_width='125px')\n)\n\n# Widgets to label the time slider with units\nTime_label = widgets.Label(value=str(Time_slide.value))\nunit_label = widgets.Label(value=str(species['timeunits'][0]))\nComposite_Time_Label = widgets.HBox([Time_label, unit_label],\n layout=widgets.Layout(width='200px'))\n\n# Labels for the parent/daughter present displays\nparent_label = widgets.Label(value=species['parent_short'][0]+' remaining',\n layout=widgets.Layout(width='200px', \n overflow_x='visible',\n overflow_y='visible') )\ndaughter_label = widgets.Label(value=species['daughter_short'][0]+' produced',\n layout=widgets.Layout(width='200px', \n overflow_x='visible',\n overflow_y='visible') )\n\n# Checkbox to choose whether to display the number of each species\n# or the fraction of each\nfrac_or_num = widgets.Checkbox(value=False, description='Display as fraction of atoms')\n\n# Widget to allow one to choose which species to work with\npick_Species = widgets.RadioButtons(options=species['parent_long'][:],\n value='Generic', description='Species:', disabled=False,\n layout=widgets.Layout(width='250px'))\n", "_____no_output_____" ], [ "# Scale for population figure\nx_sc = bq.LinearScale(min=1, max=np.sqrt(N_parent))\ny_sc = bq.LinearScale(min=1, max=np.sqrt(N_parent))\n\n# Axes for population figure\nax_x = bq.Axis(scale=x_sc, num_ticks=0)\nax_y = bq.Axis(scale=y_sc, orientation='vertical', num_ticks=0)\n\n# Creates an array of x values: [1,2,...,30,1,2,...30,.....,1,2,...,30]\nx_ls = []\nfor i in range(1,int(np.sqrt(N_parent))+1):\n x_ls.append(float(i))\nx_ls = x_ls * int(np.sqrt(N_parent))\nx_arr = np.array(x_ls)\n\n# Creates an array of y values: [1,1,...,1,2,2,...2,......,30,30,...,30]\ny_ls = []\nfor i in range(1,int(np.sqrt(N_parent))+1):\n y_ls += [float(i)] * int(np.sqrt(N_parent))\ny_arr = np.array(y_ls) \n\n# Creates a color array with the same number of entries as the number of atoms in\n# the sample\nColors = ['red'] * N_parent\n\n# Plot the population model\npopulation_scat = bq.Scatter(x=x_arr, y=y_arr, scales={'x': x_sc, 'y': y_sc}, colors =['red'])", "_____no_output_____" ], [ "# Picking a new species resets everything\npick_Species.observe(UpdateSpecies, names=['value'])\n\n# Update view from fraction to/from number\nfrac_or_num.observe(UpdateFraction, names=['value'])\n\n# Update times\nTime_slide.observe(UpdateTimes, names=['value'])\nTime_label.observe(UpdateTimes, names=['value'])\n\n# Figure for the population\nfig_population = bq.Figure(title='Population of Atoms', marks=[population_scat], axes=[ax_x, ax_y], \n background_style={'fill' : 'black'},padding_x = 0.025,\n min_aspect_ratio=1, max_aspect_ratio=1)\n\n# Boxes to organize display\nparent_box = widgets.HBox([parent_label, parent_present],\n layout=widgets.Layout(overflow_x='visible', overflow_y='visible'))\ndaughter_box = widgets.HBox([daughter_label, daughter_present],\n layout=widgets.Layout(overflow_x='visible', overflow_y='visible'))\n# Set visibility of the exact counts/fractions\nif (show_counts == False):\n parent_box.layout.visibility = 'hidden'\n daughter_box.layout.visibility = 'hidden'\n\nvalue_box = widgets.VBox([parent_box, daughter_box])\nspecies_box = widgets.HBox([value_box, pick_Species])\n\nslide_box = widgets.HBox([Time_slide, Composite_Time_Label])\nslide_check_box = widgets.VBox([slide_box, frac_or_num])\n\ntop_box = widgets.HBox([fig_counts, fig_population],\n layout=widgets.Layout(width='900px'))\ntop_box.children[0].layout.width = '450px'\ntop_box.children[1].layout.width = '450px'\n\nbottom_box = widgets.HBox([species_box, slide_check_box],\n layout=widgets.Layout(width='900px'))\nbottom_box.children[0].layout.width = '450px'\nbottom_box.children[1].layout.width = '450px'\n\n# Final display\nFinal = widgets.VBox([top_box, bottom_box])\nFinal.layout.overflow = 'hidden'\ndisplay(Final)", "_____no_output_____" ] ], [ [ "## Interactive Figure 2: Geochron Plot", "_____no_output_____" ], [ "Assuming a non-radiogenic isotope (that is, an isotope that is not the result of radioactive decay) that also will not decay, its amount should be constant. This means that for different mineral samples we can measure the ratio of parent isotope versus the non-radiogenic isotope ($P/D_i$) and daughter isotope ($D$) versus the non-radiogenic isotope ($D/D_i$) to build an geochron plot. For example, using the following isotopes\n\n- $D_i$ (non-radiogenic isotope of daughter element)\n- $D$ (Daughter Isoptope)\n- $P$ (Parent isotope)\n\nan geochron plot could plot $D/D_i$ versus $P/D_i$. \n\nWhat sets the *geochron method* (also known as the *isochron method*) apart from the just measuring parent and daughter abundances is the use of the non-radiogenic isotope of the daughter element. This avoids the assumption of no initial daughter isotope before the rock solidified (radioactive decay can occur while rock is molten).\n\nSome minerals in the rock incorporate the parent better than daughter which is why the initial amount of parent \nisotope versus daughter isotope can vary. We expect daughter versus non-radiogenic isotope ratio to be constant\nif we pick the non-radiogenic isotope to be the same element as the daughter isotope.\n\nWith all this said, it is actually often not this simple as many daughter isotopes are themselves radioactive and decay, leading to a chain of reactions, so comparing abundances of parent to daughter isotopes is not simple.\n\n*Note:* The idea for the geochron dating interactive came from a Isochron Diagram Java app at *ScienceCourseware.org*. However that app had some issues in that it didn't divide by a non-radiogenic isotope (or at least didn't mention it). In fact, they used $D_i$ for the initial amount of daughter isotope instead of the non-radiogenic isotope of the same element as the daughter isotope.", "_____no_output_____" ], [ "Consider the following questions:\n\n1. Notice that the geochron shown here is using the parent and daugther isotope ratios for 4 mineral samples in a rock. Consider the sample that starts off with the most parent isotope initially (the point initially farthest to the right). Adjust the geochron to show the samples after 1 half-life. How much of that parent isotope was there initially? How much is there after 1 half-life? Does this make sense? \n2. Consider the sample that starts off with the least parent isotope initially (the point initially farthest to the left). Adjust the geochron to show the samples after 1 half-life. How much of that parent isotope was there initially? How much is there after 1 half-life? Does this make sense? \n3. What fraction of the parent isotope should have decayed into daughter isotope after one half-life? Does it matter if the mineral sample we look at started with more parent isotope than another mineral sample in the same rock? Why or why not? \n4. Why is it as time elapses that the points representing the 4 mineral samples all move toward the upper left? Can you explain why their motions are parallel to one another?\n4. Imagine you are looking at the parent isotope Rubidium-87 which decays into Strontium-87. You examine 4 mineral samples in an meteorite and they line up in a line with a slope of 0.0666. Determine the age of that meteorite (or rather, the time since it solidified). Also explain how we can determine how much of the daughter isotope was in the meteorite initially.\n", "_____no_output_____" ] ], [ [ "##\n## Define the various isotopes we want to consider\n##\n\nisotope_info = pd.DataFrame(columns=['Name', 'PName', 'PAbbrev', 'DName', 'DAbbrev', 'DiName', 'DiAbbrev', 'HalfLife', 'HLUnits'])\nisotope_info['index'] = ['generic', 'Rb87']\nisotope_info['Name'] = ['Generic', 'Rb-87->Sr-87']\nisotope_info['PName'] = ['Parent', 'Rubidium-87']\nisotope_info['PAbbrev'] = ['P', 'Rb-87']\nisotope_info['DName'] = ['Daughter', 'Strontium-87']\nisotope_info['DAbbrev'] = ['D', 'Sr-87']\nisotope_info['DiName'] = ['Non-Radiogenic Isotope of Daughter Element', 'Strontium-86']\nisotope_info['DiAbbrev'] = ['D_i', 'Sr-86']\nisotope_info['HalfLife'] = [ 1, 48.8 ]\nisotope_info['HLUnits'] = [ 'half-lives', 'Billion years']\nisotope_info = isotope_info.set_index('index')\n\n# Set initial isotope to plot\ninit_isotope = 'generic'", "_____no_output_____" ], [ "##\n## Define the initial amounts of parent and daughter in the sample.\n##\n## In principle, I would change this depending on the isotopes we plot. But I am only plotting\n## Rb87 --> Sr-87, since that is the most classical use of this Geochron approach.\n##\n\n# Range of P to D_i fractions and initial amounts of D to D_i to consider\nP2Di_min = 0.05\nP2Di_max = 0.40\nD2Di0_min = 0.05\nD2Di0_max = 0.75\n\n# Generate three mineral samples in different thirds of the entire range\nrange_P2Di = (P2Di_max-P2Di_min)\n\n# Create sample amounts\nn_samples = 4\nnums = np.array(list(range(1, n_samples+1)))\ninitial_samples = pd.DataFrame(index=nums)\ninitial_D2Di0 = D2Di0_min + (D2Di0_max - D2Di0_min) * np.random.random()\ninitial_samples['P2Di'] = P2Di_min + (range_P2Di/n_samples) * (nums - np.random.random(n_samples))\ninitial_samples['D2Di'] = initial_D2Di0*np.ones_like(nums)\n", "_____no_output_____" ], [ "##\n## Define functions to call when building interactive plot\n##\n\ndef amt_left(sample_in, taus):\n # Generate a sample DataFrame after tau half-lifes given an initial DataFrame\n sample = sample_in.copy(deep = True)\n sample['P2Di'] = sample_in['P2Di']*((1/2)**(taus))\n sample['D2Di'] = sample_in['D2Di'] + sample_in['P2Di']*(1 - (1/2)**(taus))\n return sample\n\ndef line_points(sample):\n global x_min, x_max, y_min, y_max, initial_D2Di0\n \n # Determine the end points of a line going through the sample points.\n x_range = x_max - x_min\n y_range = y_max - y_min\n \n # Slope (extrapolate from first two points - could be done by a fit to the points)\n slope = (sample['D2Di'][2]-sample['D2Di'][1])/(sample['P2Di'][2]-sample['P2Di'][1])\n y_final = initial_D2Di0 + slope*x_range\n x_points = (x_min, x_max)\n y_points = (initial_D2Di0, y_final)\n return x_points, y_points, slope\n\ndef init2current(samples0, samples):\n # Compute the lines connecting initital and final points for plotting\n n_pts = len(samples0)\n\n xlist = []\n ylist = []\n for pt in range(1, n_pts+1):\n x = np.array([ samples0['P2Di'][pt], samples['P2Di'][pt] ])\n y = np.array([ samples0['D2Di'][pt], samples['D2Di'][pt] ])\n xlist.append(x)\n ylist.append(y)\n \n return(xlist, ylist)\n \ndef HL_changed(change):\n global isotope, sample, initial_samples, dots_current, line_current, connectors, slope_label\n \n # Determine half-life of this isotope\n idx = (isotope_info.Name == isotope.value)\n HL = float(isotope_info[idx].HalfLife.tolist()[0])\n \n # How many half-lives have passed? Use this to get new sample and line info\n this_tau = HL_slider.value / HL\n sample = amt_left(initial_samples, this_tau)\n x_sample, y_sample, slope = line_points(sample)\n \n # Update plot\n dots_current.x = sample['P2Di']\n dots_current.y = sample['D2Di']\n line_current.x = x_sample\n line_current.y = y_sample\n slope_label.value = 'Slope: {0:0.4f}'.format(slope)\n xlist, ylist = init2current(initial_samples, sample)\n connectors.x = xlist\n connectors.y = ylist\n \n \ndef isotope_changed(change):\n global ax_x_P2Di, ax_y_D2Di, HL_slider, HLlabel, UnitsText, Max_half_lives\n\n # Extract the necessary isotope descriptors from the Pandas DataFrame\n idx = (isotope_info.Name == change.new)\n HL = float(isotope_info[idx].HalfLife.tolist()[0])\n HLUnits = isotope_info[idx].HLUnits.tolist()[0]\n PAbbrev = isotope_info[idx].PAbbrev.tolist()[0]\n DAbbrev = isotope_info[idx].DAbbrev.tolist()[0]\n DiAbbrev = isotope_info[idx].DiAbbrev.tolist()[0]\n\n # Get old half-life\n idx_old = (isotope_info.Name == change.old)\n HL_old = float(isotope_info[idx_old].HalfLife.tolist()[0])\n\n # Determine current age reading from slider and adjust to new units\n init_age = HL_slider.value \n \n # Hard code generic versus others\n if (change.new != isotope_info.loc['generic'].Name):\n HL_slider.description = \"Time\"\n else: \n HL_slider.description = \"Half-lives\" \n\n # Adjust time scales\n if (HL_old < HL):\n # Adjust maximum limits first before adjusting values (since new HL > old HL)\n HL_slider.max = Max_half_lives*HL\n HLlabel.max = HL_slider.max \n HL_slider.value = HL*(init_age/HL_old)\n HLlabel.value = HL_slider.value \n else:\n # Adjust maximum limits after adjusting values (since new HL < old HL)\n HL_slider.value = HL*(init_age/HL_old)\n HLlabel.value = HL_slider.value\n HL_slider.max = Max_half_lives*HL\n HLlabel.max = HL_slider.max \n \n # Set the axes and other labels to display\n UnitsText.value = HLUnits\n ax_x_P2Di.label = '{0} / {1}'.format(PAbbrev, DiAbbrev)\n ax_y_D2Di.label = '{0} / {1}'.format(DAbbrev, DiAbbrev)\n\n", "_____no_output_____" ], [ "##\n## Set up isochron plot\n##\n\n# Largest possible fraction of decay (only go out to 5 half-lives)\nMax_half_lives = 5\nMax_decay_fraction = 1 - (1/2)**(Max_half_lives)\n\n# detemine maximum and minimum values of X and Y axes\nx_step = 0.05\nx_min = 0\nx_max = x_step * ceil(initial_samples['P2Di'][n_samples] / x_step)\ny_step = 0.04\ny_min = y_step * floor(initial_D2Di0 / y_step)\ny_max = y_step * ceil((initial_D2Di0 + initial_samples['P2Di'][n_samples] * Max_decay_fraction) / y_step)\n\n# Labels and scales for Axes\nx_P2Di = bq.LinearScale(min = x_min, max = x_max)\ny_D2Di = bq.LinearScale(min = y_min, max = y_max)\nax_x_P2Di = bq.Axis(label='P / D_i', scale=x_P2Di)\nax_y_D2Di = bq.Axis(label='D / D_i', scale=y_D2Di, orientation='vertical')\n\n# Set up initial conditions\ntaus = 0 # zero half lives past\nsample = amt_left(initial_samples, taus)\n\n##\n## Define the lines\n##\n\n# Initial amount of daughter line (with dots for initial amounts of parent)\nx_init, y_init, slope_init = line_points(initial_samples)\nline_initial = bq.Lines(x=x_init, y=y_init, scales={'x': x_P2Di, 'y': y_D2Di}, \n line_style='dashed', colors=['red'], labels=['Initial Sample'])\ndots_initial = bq.Scatter(x=initial_samples['P2Di'], y=initial_samples['D2Di'], scales={'x': x_P2Di, 'y': y_D2Di}, \n colors=['white'], stroke='red', fill= True, labels=['Initial Isochron'])\n\n# Current quantities on isochron line\nx_sample, y_sample, slope = line_points(sample)\nline_current = bq.Lines(x=x_sample, y=y_sample, scales={'x': x_P2Di, 'y': y_D2Di}, \n line_style='solid', colors=['red'], labels=['Current Isochron'])\ndots_current = bq.Scatter(x=sample['P2Di'], y=sample['D2Di'], scales={'x': x_P2Di, 'y': y_D2Di}, \n colors=['red'], stroke='red', fill= True, labels=['Current Isochron'])\n\n# Connect Initial and Current quantities on isochron line\nxlist, ylist = init2current(initial_samples, sample)\nconnectors = bq.Lines(x=xlist, y=ylist, scales={'x': x_P2Di, 'y': y_D2Di}, \n line_style='dotted', colors=['black'])\n\n##\n## Construct plot\n##\nisochron = bq.Figure(axes=[ax_x_P2Di, ax_y_D2Di], \n marks=[connectors, line_initial, dots_initial, line_current, dots_current],\n title='Geochron Diagram', \n layout={'width': '700px', 'height': '500px', \n 'max_width': '700px', 'max_height': '500px',\n 'min_width': '600px', 'min_height': '400px'})\n\n##\n## Construct controls\n##\n\n# Select Generic or Specific Isotopes\nisotope = widgets.RadioButtons(options=list(isotope_info.Name), \n value=isotope_info.loc[init_isotope].Name, description='Isotope:', \n disabled=False, \n layout=widgets.Layout(height='75px', max_height='100px', min_height='50px', \n width='200px', max_width='300px', min_width='100px'))\nisotope.observe(isotope_changed, 'value')\n\n# Slider and text field controling age\nidx = (isotope_info.Name == isotope.value)\nHL = float(isotope_info[idx].HalfLife.tolist()[0])\nHLUnits = isotope_info[idx].HLUnits.tolist()[0]\n\nHL_slider = widgets.FloatSlider(value=0, min=0, max=Max_half_lives*HL, step=0.02,\n description='Half-lives', disabled=False,\n continuous_update=False, orientation='horizontal',\n readout=False, readout_format='.2f',\n layout=widgets.Layout(height='75px', max_height='100px', min_height='50px', \n width='200px', max_width='300px', min_width='100px'))\nHL_slider.observe(HL_changed, 'value')\n\n# Get units value and units for age label, then apply them\nHLlabel = widgets.BoundedFloatText(value = HL_slider.value, min = HL_slider.min, max = HL_slider.max, \n step = HL_slider.step,\n layout={'width': '75px', 'height': '50px', \n 'max_width': '75px', 'max_height': '75px',\n 'min_width': '50px', 'min_height': '50px'})\nUnitsText = widgets.Label(value=HLUnits)\nage_label = widgets.HBox([HLlabel, UnitsText])\n# Link HL slider with this text\nwidgets.jslink((HL_slider, 'value'), (HLlabel, 'value'))\n\n# Describe slope\nslope_label = widgets.Label(value = 'Slope: {0:0.4f}'.format(slope),\n layout={'align_items':'center','align_content':'center', \n 'justify_content':'center', \n 'width': '100px', 'height': '50px', \n 'max_width': '100px', 'max_height': '75px',\n 'min_width': '50px', 'min_height': '50px'})\n\n\ncontrols = widgets.VBox( [isotope, HL_slider, age_label, slope_label], \n layout=widgets.Layout(align_content='center', align_items='center', \n justify_content='center', \n width='300px', height='500px', \n max_width='300px', max_height='500px',\n min_width='100px', min_height='400px',\n overflow_x='hidden', overflow_y='hidden') )\n\ndisplay(widgets.HBox( [isochron, controls] ) )", "_____no_output_____" ] ] ]

[ "markdown", "code", "markdown", "code" ]

[ [ "markdown", "markdown", "markdown" ], [ "code", "code", "code", "code", "code", "code", "code" ], [ "markdown", "markdown", "markdown" ], [ "code", "code", "code", "code" ] ]

[ "MIT" ]

[ "MIT" ]

[ "MIT" ]

[ [ [ "from src.modelling.bert_model import BertModel\nimport os", "_____no_output_____" ], [ "DATA_ORIGINAL = 'news'\nDATA_SUMMARIZED = 'summary'\nTEST_INDEX = 1", "_____no_output_____" ], [ "bert_model = BertModel()", "_____no_output_____" ], [ "data_path = os.path.join(os.getcwd(), r\"data\\bbc_news.csv\")\ndata = bert_model.load_data(data_path)\n\ninput_text = data[DATA_ORIGINAL][TEST_INDEX]", "_____no_output_____" ] ], [ [ "#### Load model", "_____no_output_____" ] ], [ [ "bert_model.train()", "_____no_output_____" ] ], [ [ "#### Original text", "_____no_output_____" ] ], [ [ "print(input_text)", "b'Dollar gains on Greenspan speech\\n\\nThe dollar has hit its highest level against the euro in almost three months after the Federal Reserve head said the US trade deficit is set to stabilise.\\n\\nAnd Alan Greenspan highlighted the US government\\'s willingness to curb spending and rising household savings as factors which may help to reduce it. In late trading in New York, the dollar reached $1.2871 against the euro, from $1.2974 on Thursday. Market concerns about the deficit has hit the greenback in recent months. On Friday, Federal Reserve chairman Mr Greenspan\\'s speech in London ahead of the meeting of G7 finance ministers sent the dollar higher after it had earlier tumbled on the back of worse-than-expected US jobs data. \"I think the chairman\\'s taking a much more sanguine view on the current account deficit than he\\'s taken for some time,\" said Robert Sinche, head of currency strategy at Bank of America in New York. \"He\\'s taking a longer-term view, laying out a set of conditions under which the current account deficit can improve this year and next.\"\\n\\nWorries about the deficit concerns about China do, however, remain. China\\'s currency remains pegged to the dollar and the US currency\\'s sharp falls in recent months have therefore made Chinese export prices highly competitive. But calls for a shift in Beijing\\'s policy have fallen on deaf ears, despite recent comments in a major Chinese newspaper that the \"time is ripe\" for a loosening of the peg. The G7 meeting is thought unlikely to produce any meaningful movement in Chinese policy. In the meantime, the US Federal Reserve\\'s decision on 2 February to boost interest rates by a quarter of a point - the sixth such move in as many months - has opened up a differential with European rates. The half-point window, some believe, could be enough to keep US assets looking more attractive, and could help prop up the dollar. The recent falls have partly been the result of big budget deficits, as well as the US\\'s yawning current account gap, both of which need to be funded by the buying of US bonds and assets by foreign firms and governments. The White House will announce its budget on Monday, and many commentators believe the deficit will remain at close to half a trillion dollars.\\n'\n" ] ], [ [ "#### Actual summarized text from dataset", "_____no_output_____" ] ], [ [ "actual_summary = data[DATA_SUMMARIZED][TEST_INDEX]\nprint(actual_summary)", "b'The dollar has hit its highest level against the euro in almost three months after the Federal Reserve head said the US trade deficit is set to stabilise.China\\'s currency remains pegged to the dollar and the US currency\\'s sharp falls in recent months have therefore made Chinese export prices highly competitive.Market concerns about the deficit has hit the greenback in recent months.\"I think the chairman\\'s taking a much more sanguine view on the current account deficit than he\\'s taken for some time,\" said Robert Sinche, head of currency strategy at Bank of America in New York.The recent falls have partly been the result of big budget deficits, as well as the US\\'s yawning current account gap, both of which need to be funded by the buying of US bonds and assets by foreign firms and governments.\"He\\'s taking a longer-term view, laying out a set of conditions under which the current account deficit can improve this year and next.\"'\n" ] ], [ [ "#### Summarized text", "_____no_output_____" ] ], [ [ "summary = bert_model.predict(input_text)\nprint(summary)", "b'Dollar gains on Greenspan speech\\n\\nThe dollar has hit its highest level against the euro in almost three months after the Federal Reserve head said the US trade deficit is set to stabilise.\\n\\nAnd Alan Greenspan highlighted the US government\\'s willingness to curb spending and rising household savings as factors which may help to reduce it. I think the chairman\\'s taking a much more sanguine view on the current account deficit than he\\'s taken for some time,\" said Robert Sinche, head of currency strategy at Bank of America in New York. \" But calls for a shift in Beijing\\'s policy have fallen on deaf ears, despite recent comments in a major Chinese newspaper that the \"time is ripe\" for a loosening of the peg.\n" ] ], [ [ "#### Evaluation", "_____no_output_____" ] ], [ [ "reference_data = data[DATA_ORIGINAL].sample(n=10, random_state=42)\nreference_data", "_____no_output_____" ], [ "# candidate_data = reference_data.apply(lambda x: bert_model.predict(x))\ncandidate_data = reference_data.map(bert_model.predict)\ncandidate_data", "_____no_output_____" ] ], [ [ "#### Score for 10 samples", "_____no_output_____" ] ], [ [ "precision, recall, f1 = bert_model.evaluation(preds=candidate_data, refs=reference_data)\nprint(f'Precision: {precision}')\nprint(f'Recall: {recall}')\nprint(f'F1: {f1}')", "calculating scores...\ncomputing bert embedding.\n" ] ], [ [ "#### Score for 100 samples", "_____no_output_____" ] ], [ [ "precision, recall, f1 = bert_model.evaluation(preds=preds, refs=refs)\nprint(f'Precision: {precision}')\nprint(f'Recall: {recall}')\nprint(f'F1: {f1}')", "calculating scores...\ncomputing bert embedding.\n" ] ] ]

[ "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code" ]

[ [ "code", "code", "code", "code" ], [ "markdown" ], [ "code" ], [ "markdown" ], [ "code" ], [ "markdown" ], [ "code" ], [ "markdown" ], [ "code" ], [ "markdown" ], [ "code", "code" ], [ "markdown" ], [ "code" ], [ "markdown" ], [ "code" ] ]

[ "BSD-3-Clause" ]

[ "BSD-3-Clause" ]

[ "BSD-3-Clause" ]

[ [ [ "# Execution\n\nThis notebook explains how PartiQL executes, with some discussion of its implementation.", "_____no_output_____" ], [ "## Data model\n\nAs discussed in [02-data-model.ipynb](02-data-model.ipynb), the only allowed types are PLURP (PLUR in this demo). In a rowwise, dynamically typed interpreter, this means there are only values, records, and lists. Since PartiQL is list order-independent and merges items with the same key, the lists are really sets, though the word \"list\" appears in the implementation and error messages.\n\nEvery value entering or exiting the PartiQL interpreter is a `data.Instance`, even simple numbers.", "_____no_output_____" ] ], [ [ "import data\n\narrays = data.RecordArray({\n \"x\": data.PrimitiveArray([0.1, 0.2, 0.3]),\n \"y\": data.PrimitiveArray([1000, 2000, 3000]),\n \"table1\": data.ListArray([0, 3, 3], [3, 3, 5], data.RecordArray({\n \"a\": data.PrimitiveArray([1.1, 2.2, 3.3, 4.4, 5.5]),\n \"b\": data.PrimitiveArray([100, 200, 300, 400, 500])\n })\n ),\n \"table2\": data.ListArray([0, 2, 3], [2, 3, 8], data.RecordArray({\n \"b\": data.PrimitiveArray([True, False, True, False, False, True, True, True]),\n \"c\": data.PrimitiveArray([10, 20, 100, 1, 2, 3, 4, 5])\n })\n ),\n \"stuff\": data.ListArray([0, 3, 3], [3, 3, 5], data.PrimitiveArray([1, 2, 3, 4, 5]))\n})\n\narrays.tolist()", "_____no_output_____" ], [ "arrays.setindex()\n\ninstances = data.instantiate(arrays)\ninstances", "_____no_output_____" ] ], [ [ "## Inputs and outputs\n\nNext, we'll run a few simple examples to show what the runtime engine requires and produces.\n\nIf you have not already done so, install the [Lark parser](https://github.com/lark-parser/lark#readme) and Matplotlib.", "_____no_output_____" ] ], [ [ "!pip install lark-parser matplotlib", "Requirement already satisfied: lark-parser in /home/jpivarski/miniconda3/lib/python3.7/site-packages (0.7.1)\nRequirement already satisfied: matplotlib in /home/jpivarski/miniconda3/lib/python3.7/site-packages (3.1.1)\nRequirement already satisfied: cycler>=0.10 in /home/jpivarski/miniconda3/lib/python3.7/site-packages (from matplotlib) (0.10.0)\nRequirement already satisfied: pyparsing!=2.0.4,!=2.1.2,!=2.1.6,>=2.0.1 in /home/jpivarski/miniconda3/lib/python3.7/site-packages (from matplotlib) (2.4.0)\nRequirement already satisfied: python-dateutil>=2.1 in /home/jpivarski/miniconda3/lib/python3.7/site-packages (from matplotlib) (2.8.0)\nRequirement already satisfied: numpy>=1.11 in /home/jpivarski/miniconda3/lib/python3.7/site-packages (from matplotlib) (1.16.4)\nRequirement already satisfied: kiwisolver>=1.0.1 in /home/jpivarski/miniconda3/lib/python3.7/site-packages (from matplotlib) (1.1.0)\nRequirement already satisfied: six in /home/jpivarski/miniconda3/lib/python3.7/site-packages (from cycler>=0.10->matplotlib) (1.12.0)\nRequirement already satisfied: setuptools in /home/jpivarski/miniconda3/lib/python3.7/site-packages (from kiwisolver>=1.0.1->matplotlib) (41.0.1)\n" ], [ "import interpreter", "_____no_output_____" ] ], [ [ "The interpreter takes a source string and a `data.ListInstance` of `data.RecordInstances` as input.\n\nIt returns any newly assigned variables (as a `data.ListInstance` of `data.RecordInstances`) and the hierarchy of counters, which may be a single counter (total number of events) or a directory of histograms.", "_____no_output_____" ] ], [ [ "output, counter = interpreter.run(r\"\"\"\n\nz = x + 1\n\nhist z by regular(100, 1, 1.5) named \"h\"\n\n\"\"\", instances)", "_____no_output_____" ], [ "output", "_____no_output_____" ], [ "counter.allkeys()", "_____no_output_____" ], [ "counter[\"h\"].mpl()", "_____no_output_____" ] ], [ [ "Variables assigned in nested blocks will *not* be returned, but histograms defined in these blocks will.", "_____no_output_____" ] ], [ [ "output, counter = interpreter.run(r\"\"\"\n\ntable1 with {\n z = a + 1\n hist z by regular(100, 0, 10) named \"h\"\n}\n\n\"\"\", instances)", "_____no_output_____" ], [ "output", "_____no_output_____" ], [ "counter[\"h\"].mpl()", "_____no_output_____" ] ], [ [ "To get nested quantities as output, be sure to assign them to a top-level variable.", "_____no_output_____" ] ], [ [ "output, counter = interpreter.run(r\"\"\"\n\ntop = table1 with {\n z = a + 1\n hist z by regular(100, 0, 10) named \"h\"\n}\n\n\"\"\", instances)", "_____no_output_____" ], [ "output", "_____no_output_____" ] ], [ [ "Variables assigned at the top-level of the source are a functional return value, but histograms and counters are a side-effect.\n\nHistograms nested in a `cut` or `vary` are nested in their counters.", "_____no_output_____" ] ], [ [ "output, counter = interpreter.run(r\"\"\"\ncut x > 0.1 named \"c\" {\n table2 with {\n hist c by regular(100, 0, 10) named \"h\"\n }\n}\n\"\"\", instances)", "_____no_output_____" ], [ "counter.allkeys()", "_____no_output_____" ], [ "counter[\"c\"]", "_____no_output_____" ], [ "counter[\"c/h\"].mpl()", "_____no_output_____" ] ], [ [ "## Scalar expressions", "_____no_output_____" ], [ "Trivial assignment is one way to pass on values from intput to output.", "_____no_output_____" ] ], [ [ "output, counter = interpreter.run(r\"\"\"\n\nx = x\n\n\"\"\", instances); output", "_____no_output_____" ] ], [ [ "Missing values are handled as unions of records with a field with records without that field. They're only passed through an expression if safe navigation operators (`?`) are used (or explicit checks for `has`).", "_____no_output_____" ] ], [ [ "output, counter = interpreter.run(r\"\"\"\n\na = if x > 0.1 then 100\n\nb = if x > 0.1 then 100 else 200\n\nc = ?a\n\nd = if has a then 100 else 200\n\n\"\"\", instances); output", "_____no_output_____" ] ], [ [ "Temporary variables don't appear if they're in a curly-brackets scope.", "_____no_output_____" ] ], [ [ "output, counter = interpreter.run(r\"\"\"\n\na = {\n tmp1 = x + 1\n tmp2 = x * 100\n tmp2**2\n}\n\n\"\"\", instances); output", "_____no_output_____" ] ], [ [ "## Set operations", "_____no_output_____" ], [ "The `as` operator turns any set into a set of records nested within a single field.\n\nIn the example below, `out1` does not have a record structure, but `out2` does—the same data is packed in a field named `nested`.", "_____no_output_____" ] ], [ [ "output, counter = interpreter.run(r\"\"\"\n\nout1 = stuff\nout2 = stuff as nested\n\n\"\"\", instances); output", "_____no_output_____" ] ], [ [ "It is also a special case of sampling without replacement.\n\nBelow, we see that each record has an `x` and a `y` field, and these are unique pairs of the original data, per-event: `(1, 2)`, `(1, 3)`, `(2, 3)` in the first event and only `(4, 5)` in the third event. (The second event is empty.)", "_____no_output_____" ] ], [ [ "output, counter = interpreter.run(r\"\"\"\n\nout = stuff as (x, y)\n\n\"\"\", instances); output", "_____no_output_____" ] ], [ [ "The natural alternative to sampling without replacement is sampling with replacement, which can be built from `as` and `cross` (cross-join).\n\nThe cross-join computes a Cartesian product, just as it does in SQL. Promoting each set of integers into sets of records makes them mergable (`x` and `y` go into the same output records).\n\nNow the pairs are `(1, 1)`, `(1, 2)`, `(1, 3)`, `(2, 1)`, `(2, 2)`, `(2, 3)`, `(3, 1)`, `(3, 2)`, `(3, 3)` in the first event and `(4, 4)`, `(4, 5)`, `(5, 4)`, `(5, 5)` in the third event. (The second event is empty.)", "_____no_output_____" ] ], [ [ "output, counter = interpreter.run(r\"\"\"\n\nout = stuff as x cross stuff as y\n\n\"\"\", instances); output", "_____no_output_____" ] ], [ [ "For added flair, we can `group by` either `x` or `y` to build sets of sets. The Cartesian grid pattern should be more clear.", "_____no_output_____" ] ], [ [ "output, counter = interpreter.run(r\"\"\"\n\nout = stuff as x cross stuff as y group by x\n\n\"\"\", instances); output", "_____no_output_____" ] ], [ [ "The cross-join/Cartesian product is one kind of a join: two sets of records are combined to make one set of records. The output records have fields from both input record types and the set is derived from a combinatorial rule, in this case: \"for all in the left set, merge with each of the right set.\"\n\nIf we didn't give the records in the left and right sets distinct field names with `as`, then they could overshadow each other. That is, if the left and right both had a field named `x`, only the left's `x` would show up in the results (by convention, left wins over right). Perhaps it would be better if a situation like this raises an error...\n\nThe output records don't necessarily have distinct values for all fields. The entities, defined by hidden surrogate keys, are guaranteed distinct, but the field values might not be.", "_____no_output_____" ] ], [ [ "output, counter = interpreter.run(r\"\"\"\n\nout = stuff as x cross stuff as x\n\n\"\"\", instances); output", "_____no_output_____" ] ], [ [ "In the above, it looks like these sets contain multiple values of `{\"x\": 4}` and `{\"x\": 5}`, but their index keys are different. (They are different ordered tuples of elements from the left and right sets.)", "_____no_output_____" ] ], [ [ "[[y.row for y in x[\"out\"]] for x in output]", "_____no_output_____" ] ], [ [ "Cross-join is, in a sense, the simplest join because it creates a new index: `CrossRef(<Ref X>, <Ref Y>)` is distinct from `<Ref X>` and `<Ref Y>`, and the only way to get another one is to `cross` sets with indexes `<Ref X>` and `<Ref Y>` again.\n\nThe next-simplest is `join`, which requires the left and right sets to have the same index and returns a set with the same index. In SQL terms, PartiQL's `join` is an `INNER JOIN`, taking an element from the left and right sets only if they have the same index key. Again in SQL terms, this `INNER JOIN` is implicitly `ON` the hidden surrogate index, not any other columns. With that restriction, `join` effectively becomes set intersection.\n\nI had considered calling it `intersect`, rather than `join` because it is set intersection, but the way fields from records in the left and right sets are mixed is more reminiscent of an SQL join.", "_____no_output_____" ] ], [ [ "output, counter = interpreter.run(r\"\"\"\n\nout = stuff as x join stuff as y\n\n\"\"\", instances); output", "_____no_output_____" ] ], [ [ "To show what this is doing, let's filter the left and right sets with `where`. Even and odd values have no overlap.", "_____no_output_____" ] ], [ [ "output, counter = interpreter.run(r\"\"\"\n\nevens = stuff as x where x % 2 == 0\nodds = stuff as x where x % 2 == 1\n\nout = evens join odds\n\n\"\"\", instances); output", "_____no_output_____" ] ], [ [ "But the overlap of evens and `2 <= x and x <= 4` is `2` and `4`.", "_____no_output_____" ] ], [ [ "output, counter = interpreter.run(r\"\"\"\n\nevens = stuff as x where x % 2 == 0\nmiddle = stuff as x where 2 <= x and x <= 4\n\nout = evens join middle\n\n\"\"\", instances); output", "_____no_output_____" ] ], [ [ "To see how this is like an SQL `INNER JOIN`, consider records with different fields: the left records have an `x` and the right records have a `y`. When they have an overlapping index, the merged records have `x` and `y`.", "_____no_output_____" ] ], [ [ "output, counter = interpreter.run(r\"\"\"\n\nevens = stuff as x where x % 2 == 0\nmiddle = stuff as y where 2 <= y and y <= 4\n\nout = evens join middle\n\n\"\"\", instances); output", "_____no_output_____" ] ], [ [ "The opposite of PartiQL's `join` (like SQL's `INNER JOIN` on the hidden surrogate key) is PartiQL's `union` (like SQL's `FULL OUTER JOIN` on the hidden surrogate key).\n\nJust as `join` acts as set intersection, `union` acts as set union. I chose to use the word \"`union`\" in this case because field names of the records in the left and right sets are often different, so this behaves intuitively as a user would expect set union to behave.\n\nHere is an example where the field name is the same, but the left and right sets are non-overlapping (so the union is just their concatenation).", "_____no_output_____" ] ], [ [ "output, counter = interpreter.run(r\"\"\"\n\nevens = stuff as x where x % 2 == 0\nodds = stuff as x where x % 2 == 1\n\nout = evens union odds\n\n\"\"\", instances); output", "_____no_output_____" ] ], [ [ "Here's the same example, but with fields named `x` and `y`. If the output has a field named `x`, you know it came from the left, and if it has a field named `y`, you know it came from the right.", "_____no_output_____" ] ], [ [ "output, counter = interpreter.run(r\"\"\"\n\nevens = stuff as x where x % 2 == 0\nodds = stuff as y where y % 2 == 1\n\nout = evens union odds\n\n\"\"\", instances); output", "_____no_output_____" ] ], [ [ "Often, that would be used to combine particle lists, like `leptons = electrons union muons`. In the example below, we'll take a union of `table1` and `table2`.", "_____no_output_____" ] ], [ [ "output, counter = interpreter.run(r\"\"\"\n\nout = table1 union table2\n\n\"\"\", instances); output", "_____no_output_____" ] ], [ [ "The `table1` and `table2` sets both have a `b` field, so `b` in the output records might have come from `table1` (in which case, it's an integer) or might have come from `table2` (in which case, it's boolean). To make the provenance clear, we can use `as` to label them.\n\nBelow, the output records either have a `left` field or they have a `right` field.", "_____no_output_____" ] ], [ [ "output, counter = interpreter.run(r\"\"\"\n\nout = table1 as left union table2 as right\n\n\"\"\", instances); output", "_____no_output_____" ] ], [ [ "Also notice that the index is a union of the two input indexes. Keys either come from `<Ref 1>` or `<Ref 2>`.", "_____no_output_____" ] ], [ [ "[[y.row for y in x[\"out\"]] for x in output]", "_____no_output_____" ] ], [ [ "So `cross` creates a new index reference (`CrossRef(<Ref X>, <Ref Y>)`), `join` returns the input index references (finding the ones that are in both left and right), and `union` creates a union of keys out of the left and right index references.\n\nThere's one more set operation: `except`. This is a set difference, removing elements from the left set of records whose keys can be found in the right set of records.", "_____no_output_____" ] ], [ [ "output, counter = interpreter.run(r\"\"\"\n\nevens = stuff as x where x % 2 == 0\n\nout = stuff as x except evens\n\n\"\"\", instances); output", "_____no_output_____" ] ], [ [ "I could have implemented a symmetric set difference or an equivalent of SQL's `LEFT OUTER JOIN` and `RIGHT OUTER JOIN`, but they're straightforward extensions of what I've implemented here (and it's not clear they would be needed for a physics analysis).", "_____no_output_____" ], [ "### Equality and set membership\n\nIt should be emphasized that the set operations defined above use a definition of equality that only requires two objects' index keys to be the same. They can have different field names and different values of their fields and be considered mergable by `join` and they can have the same field names and the same field values and be considered different elements by `union`.\n\nThis is useful because a user might want to enrich a subset of particles with new computed values and then mix in particles from the original set. The \"entity identity\" that must be maintained is the *particles,* not any attached values. To illustrate this, let's do some work on a subset of `table2`, different work on another subset of `table2`, and then bring them together for the final result.\n\nThis is an example of a DAG, described in the abstract in [02-data-model.ipynb#Joins-and-index-compatibility](02-data-model.ipynb#Joins-and-index-compatibility).", "_____no_output_____" ] ], [ [ "output, counter = interpreter.run(r\"\"\"\nout = {\n subset1 = table2 where c % 2 == 1 with {\n z = 2*c\n }\n subset2 = table2 where not b with {\n z = 10*c\n }\n subset1 union subset2\n}\n\"\"\", instances); output", "_____no_output_____" ] ], [ [ "Note that this is not how simple equality—tested by the `==` operator—should work. Simple equality should return `True` if the left and right are the same value or have the same field names and field values, irrespective of whether they have the same indexes. The same applies to `!=`, `in`, and `not in`, which are all scalar expressions.", "_____no_output_____" ] ], [ [ "output, counter = interpreter.run(r\"\"\"\n\nout = 2 in stuff\n\n\"\"\", instances); output", "_____no_output_____" ] ], [ [ "## Modifying sets of records\n\nSQL's `SELECT` statement lets you create, replace, and remove columns from a table without changing its index (except inasmuch as a visible or natural index is a column that can be changed like any other). The equivalents in PartiQL are:\n\n * `with { ... }` to create or replace fields to a set of records (copying over any fields not specified),\n * `to { ... }` to replace all fields in a set of records (only copying a field if `x = x` is specified),\n * `to ...` to replace the set of records with the result of an expression, which might not even be a record.\n\nHere are a few examples.", "_____no_output_____" ] ], [ [ "output, counter = interpreter.run(r\"\"\"\n\nout = table1 with {\n c = 10*a # add c as a new field\n}\n\n\"\"\", instances); output", "_____no_output_____" ], [ "output, counter = interpreter.run(r\"\"\"\n\nout = table1 to {\n c = 10*a # c is the only field in a new set of records\n}\n\n\"\"\", instances); output", "_____no_output_____" ], [ "output, counter = interpreter.run(r\"\"\"\n\nout = table1 to 10*a # the results are simple values, not records (in this case, numbers)\n\n\"\"\", instances); output", "_____no_output_____" ] ], [ [ "## Group-by\n\nThe `GROUP BY` operator is a major part of SQL: it changes a `SELECT` operation from a filtered transformation of a table into an aggregation over subtables, each defined by a distinct value of the expression following `GROUP BY`. It changes the nature of the whole query—expressions after `SELECT` must be reducers (converting lists/sets into scalars).\n\nIn PartiQL, `group by` acts as a simple function that converts sets of records into sets of sets of those records, grouped by distinct values of the expression following `group by`. If you decide to aggregate those inner sets, that's your choice.", "_____no_output_____" ] ], [ [ "output, counter = interpreter.run(r\"\"\"\n\ngrouped = table2 group by b\n\n\"\"\", instances); output", "_____no_output_____" ] ], [ [ "Below, we aggregate over the inner sets by modifying the set of records as we would modify any other set of records—by defining new fields in a `to` block. The aggregation functions (`count`, `sum`, `min`, `max`, `any`, `all`) operate on sets of values.\n\nIf we wanted to aggregate `b` (because we grouped by `b`; it would have the same value in every element) without aggregating `c`, we could do that. The data model is capable of managing sets of records whose fields are a scalar and a set.", "_____no_output_____" ] ], [ [ "output, counter = interpreter.run(r\"\"\"\n\naggregated = table2 group by b as grouped to {\n b = any(grouped.b)\n c = sum(grouped.c)\n}\n\n\"\"\", instances); output", "_____no_output_____" ] ], [ [ "## Min by and max by\n\nWhereas `group by` restructures a set of records into a set of sets of records, `min by` and `max by` restructure a set of records into a single record. If the set is empty, this returns missing data.", "_____no_output_____" ] ], [ [ "output, counter = interpreter.run(r\"\"\"\n\nout = table1 max by a\n\n\"\"\", instances); output", "_____no_output_____" ], [ "output, counter = interpreter.run(r\"\"\"\n\nout = (table1 max by a).b\n\n\"\"\", instances); output", "_____no_output_____" ] ], [ [ "# Concluding remarks\n\nSQL has broad applicability because it is a complete system for transforming sets without the complexity of a generic programming language. Unlike functional programming, which is an implementation of Alonzo Church's Lambda Calculus (Turing complete), SQL is an implementation of Edgar Codd's [Relational Algebra](https://en.wikipedia.org/wiki/Relational_algebra), a more restricted but still useful system. In principle, a Turing complete language must be executed to find out what it does, but set operations can be analyzed analytically.\n\nSQL's claims to \"completeness\" rest on its implementation of *almost* every operation in the Relational Algebra. PartiQL has the same features:\n\n * Cartesian product: `cross`,\n * projection: `to` (without curly brackets),\n * selection: `with` or `to` (with curly brackets),\n * rename: `to` (with curly brackets),\n * inner and outer join: `join` and `union`, which double as intersection and union, given our choice of surrogate index,\n * θ-join: `join` with `where`,\n * semijoin: (not implemented; this is the `LEFT OUTER JOIN` and `RIGHT OUTER JOIN` of SQL, discussed above),\n * antijoin: `except`,\n * division: (not implemented, but not implemented in SQL, either),\n\nFurthermore, PartiQL uses true sets as containers, not bags (sets with duplicates). Unlike SQL, PartiQL has prescribed indexes and doesn't let the user choose new indexes for `JOIN` operations, but this is because entity identity is defined by reconstruction and does not have to be discovered from natural keys in the data.\n\nLike SQL, PartiQL has aggregation (reducer functions) and `group by`, both of which are extensions beyond Relational Algebra.\n\nSQL's only container type is a table, which is a bag of relations, but PartiQL handles sets of values, relations (\"records\" in PartiQL), as well as sets. This can be modeled in SQL using multiple tables and foreign keys, but PartiQL hides the keys and presents conventional data structures.\n\nAdditionally, PartiQL is capable of all the `JaggedArray` operations that have been needed in awkward-array, except for those dealing with order and forcing arrays to have a fixed size (e.g. `JaggedArray.pad`).\n\nThe pattern matching of my May 2019 language is expressible in PartiQL, though without the nicety of a pattern-match syntax. There could be an elegant way to merge the two, though PartiQL's paradigm of naming samples with and without replacement before building any structures is more powerful than the pattern-matching syntax. (There are long-distance constraints that can't be expressed in pattern-matching, but can with PartiQL.)\n\nDifficult combinatorics, such as [Benchmark 8](https://github.com/iris-hep/adl-benchmarks-index#functionality-benchmarks), appear to be trivial and quite readable in this language.\n\nThus, I conclude that a language built out of\n\n 1. SQL-style surrogate keys (hidden),\n 2. SQL-style joins (on the hidden keys),\n 3. with a deep structure-friendly data model and a more composable syntax than SQL\n\nwould be a useful language for doing particle physics, would be different enough in capabilities to provide more \"value added,\" and would be more tightly controlled (more declarative, more optimizable) than a general programming language, including functional languages like Spark and LINQ.", "_____no_output_____" ] ] ]

[ "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown" ]

[ [ "markdown", "markdown" ], [ "code", "code" ], [ "markdown" ], [ "code", "code" ], [ "markdown" ], [ "code", "code", "code", "code" ], [ "markdown" ], [ "code", "code", "code" ], [ "markdown" ], [ "code", "code" ], [ "markdown" ], [ "code", "code", "code", "code" ], [ "markdown", "markdown" ], [ "code" ], [ "markdown" ], [ "code" ], [ "markdown" ], [ "code" ], [ "markdown", "markdown" ], [ "code" ], [ "markdown" ], [ "code" ], [ "markdown" ], [ "code" ], [ "markdown" ], [ "code" ], [ "markdown" ], [ "code" ], [ "markdown" ], [ "code" ], [ "markdown" ], [ "code" ], [ "markdown" ], [ "code" ], [ "markdown" ], [ "code" ], [ "markdown" ], [ "code" ], [ "markdown" ], [ "code" ], [ "markdown" ], [ "code" ], [ "markdown" ], [ "code" ], [ "markdown" ], [ "code" ], [ "markdown" ], [ "code" ], [ "markdown" ], [ "code" ], [ "markdown", "markdown" ], [ "code" ], [ "markdown" ], [ "code" ], [ "markdown" ], [ "code", "code", "code" ], [ "markdown" ], [ "code" ], [ "markdown" ], [ "code" ], [ "markdown" ], [ "code", "code" ], [ "markdown" ] ]

[ "MIT" ]

[ "MIT" ]

[ "MIT" ]

[ [ [ "# Load the Dataset", "_____no_output_____" ] ], [ [ "import pandas as pd\nimport numpy as np\nimport random\nurl = \"https://archive.ics.uci.edu/ml/machine-learning-databases/iris/iris.data\"\niris = pd.read_csv(url, \n header=None,\n names = ['sepal_length', \n 'sepal_width', \n 'petal_length', \n 'petal_width', \n 'species'])", "_____no_output_____" ], [ "## Definitions of classifying classes\nclasses = list(pd.unique(iris.species))\nnumClasses = len(classes)", "_____no_output_____" ] ], [ [ "## Extracting the Feature Matrix", "_____no_output_____" ] ], [ [ "X = np.matrix(iris.iloc[:, 0:4])\nX = X.astype(np.float)\nm, n = X.shape", "_____no_output_____" ] ], [ [ "## Extracting the response", "_____no_output_____" ] ], [ [ "y = np.asarray(iris.species)", "_____no_output_____" ] ], [ [ "## Extracting features for different classes", "_____no_output_____" ] ], [ [ "CLS = []\nfor each in classes:\n CLS.append(np.matrix(iris[iris.species == each].iloc[:, 0:4]))", "_____no_output_____" ], [ "len(CLS)", "_____no_output_____" ] ], [ [ "## The real meat", "_____no_output_____" ], [ "#### Calculating the mean and variance of each features for each class", "_____no_output_____" ] ], [ [ "pArray = []\ndef calculate_mean_and_variance(CLS, n, numClasses):\n for i in range(numClasses):\n pArray.append([])\n for x in range(n):\n mean = np.mean(CLS[i][:, x])\n var = np.var(CLS[i][:, x])\n pArray[i].append([mean, var])", "_____no_output_____" ], [ "calculate_mean_and_variance(CLS, n, numClasses)\nfor each in pArray:\n print(each, end='\\n\\n')", "[[5.0060000000000002, 0.12176400000000002], [3.4180000000000001, 0.14227600000000001], [1.464, 0.029504000000000002], [0.24399999999999999, 0.011264000000000003]]\n\n[[5.9359999999999999, 0.261104], [2.7700000000000005, 0.096500000000000016], [4.2599999999999998, 0.21640000000000004], [1.3259999999999998, 0.038323999999999997]]\n\n[[6.5879999999999983, 0.39625600000000011], [2.9740000000000002, 0.10192399999999999], [5.5520000000000005, 0.29849600000000004], [2.0260000000000002, 0.07392399999999999]]\n\n" ] ], [ [ "#### Choosing training dataset (Random Choosing)", "_____no_output_____" ] ], [ [ "# Choosing 70% of the dataset for training Randomly\nrandom_index = random.sample(range(m), int(m * 0.7))", "_____no_output_____" ], [ "def probability(mean, stdev, x):\n ", "_____no_output_____" ] ], [ [ "#### Creating the actual Baysean Classifier", "_____no_output_____" ] ], [ [ "def classify_baysean():\n correct_predictions = 0\n for index in random_index:\n result = []\n x = X[index, :]\n for eachClass in range(numClasses):\n result.append([])\n prior = 1 / numClasses\n \n # For sepal_length\n prosterior_feature_1 = probability(pArray[index][0][0], \n pArray[index][0][1], \n x[0])\n # For sepal_width\n prosterior_feature_2 = probability(pArray[index][1][0],\n pArray[index][1][1],\n x[1])\n # For petal_length\n prosterior_feature_3 = probability(pArray[index][2][0],\n pArray[index][2][1],\n x[2])\n # For petal_width\n prosterior_feature_4 = probability(pArray[index][3][0],\n pArray[index][3][1],\n x[3])\n joint = prosterior_feature_1 * prosterior_feature_2 * \\\n prosterior_feature_3 * prosterior_feature_4 * prior\n result[index].append(joint)\n print(result[index])", "_____no_output_____" ], [ "classify_baysean()", "_____no_output_____" ], [ "x = X[49,:][:, 0]", "_____no_output_____" ], [ "x", "_____no_output_____" ], [ "import math", "_____no_output_____" ], [ "mean = pArray[0][0][0]\nstdev = pArray[0][0][1]\nexponent = math.exp(-(math.pow(x-mean,2)/(2*stdev)))", "_____no_output_____" ], [ "exponent", "_____no_output_____" ], [ "a = [[1, 2, 3, 4], [5, 6, 7, 8]]", "_____no_output_____" ], [ "for attribute in zip(*a):\n print(attribute)", "(1, 5)\n(2, 6)\n(3, 7)\n(4, 8)\n" ], [ "df = pd.DataFrame(np.random.randn(100, 2))", "_____no_output_____" ], [ "df", "_____no_output_____" ], [ "msk = np.random.rand(len(df)) < 0.8", "_____no_output_____" ], [ "msk", "_____no_output_____" ] ] ]

[ "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code" ]

[ [ "markdown" ], [ "code", "code" ], [ "markdown" ], [ "code" ], [ "markdown" ], [ "code" ], [ "markdown" ], [ "code", "code" ], [ "markdown", "markdown" ], [ "code", "code" ], [ "markdown" ], [ "code", "code" ], [ "markdown" ], [ "code", "code", "code", "code", "code", "code", "code", "code", "code", "code", "code", "code", "code" ] ]

[ "MIT" ]

[ "MIT" ]

[ "MIT" ]

[ [ [ "# Optimization Methods\n\nUntil now, you've always used Gradient Descent to update the parameters and minimize the cost. In this notebook, you will learn more advanced optimization methods that can speed up learning and perhaps even get you to a better final value for the cost function. Having a good optimization algorithm can be the difference between waiting days vs. just a few hours to get a good result. \n\nGradient descent goes \"downhill\" on a cost function $J$. Think of it as trying to do this: \n<img src=\"images/cost.jpg\" style=\"width:650px;height:300px;\">\n<caption><center> <u> **Figure 1** </u>: **Minimizing the cost is like finding the lowest point in a hilly landscape**<br> At each step of the training, you update your parameters following a certain direction to try to get to the lowest possible point. </center></caption>\n\n**Notations**: As usual, $\\frac{\\partial J}{\\partial a } = $ `da` for any variable `a`.\n\nTo get started, run the following code to import the libraries you will need.", "_____no_output_____" ] ], [ [ "import numpy as np\nimport matplotlib.pyplot as plt\nimport scipy.io\nimport math\nimport sklearn\nimport sklearn.datasets\n\nfrom opt_utils import load_params_and_grads, initialize_parameters, forward_propagation, backward_propagation\nfrom opt_utils import compute_cost, predict, predict_dec, plot_decision_boundary, load_dataset\nfrom testCases import *\n\n%matplotlib inline\nplt.rcParams['figure.figsize'] = (7.0, 4.0) # set default size of plots\nplt.rcParams['image.interpolation'] = 'nearest'\nplt.rcParams['image.cmap'] = 'gray'", "C:\\Users\\Theochem\\Desktop\\DeepLearning\\deep-learning-coursera-master\\Improving Deep Neural Networks Hyperparameter tuning, Regularization and Optimization\\opt_utils.py:76: SyntaxWarning: assertion is always true, perhaps remove parentheses?\n assert(parameters['W' + str(l)].shape == layer_dims[l], layer_dims[l-1])\nC:\\Users\\Theochem\\Desktop\\DeepLearning\\deep-learning-coursera-master\\Improving Deep Neural Networks Hyperparameter tuning, Regularization and Optimization\\opt_utils.py:77: SyntaxWarning: assertion is always true, perhaps remove parentheses?\n assert(parameters['W' + str(l)].shape == layer_dims[l], 1)\nC:\\Users\\Theochem\\Anaconda3\\lib\\site-packages\\h5py\\__init__.py:36: FutureWarning: Conversion of the second argument of issubdtype from `float` to `np.floating` is deprecated. In future, it will be treated as `np.float64 == np.dtype(float).type`.\n from ._conv import register_converters as _register_converters\n" ] ], [ [ "## 1 - Gradient Descent\n\nA simple optimization method in machine learning is gradient descent (GD). When you take gradient steps with respect to all $m$ examples on each step, it is also called Batch Gradient Descent. \n\n**Warm-up exercise**: Implement the gradient descent update rule. The gradient descent rule is, for $l = 1, ..., L$: \n$$ W^{[l]} = W^{[l]} - \\alpha \\text{ } dW^{[l]} \\tag{1}$$\n$$ b^{[l]} = b^{[l]} - \\alpha \\text{ } db^{[l]} \\tag{2}$$\n\nwhere L is the number of layers and $\\alpha$ is the learning rate. All parameters should be stored in the `parameters` dictionary. Note that the iterator `l` starts at 0 in the `for` loop while the first parameters are $W^{[1]}$ and $b^{[1]}$. You need to shift `l` to `l+1` when coding.", "_____no_output_____" ] ], [ [ "# GRADED FUNCTION: update_parameters_with_gd\n\ndef update_parameters_with_gd(parameters, grads, learning_rate):\n \"\"\"\n Update parameters using one step of gradient descent\n \n Arguments:\n parameters -- python dictionary containing your parameters to be updated:\n parameters['W' + str(l)] = Wl\n parameters['b' + str(l)] = bl\n grads -- python dictionary containing your gradients to update each parameters:\n grads['dW' + str(l)] = dWl\n grads['db' + str(l)] = dbl\n learning_rate -- the learning rate, scalar.\n \n Returns:\n parameters -- python dictionary containing your updated parameters \n \"\"\"\n\n L = len(parameters) // 2 # number of layers in the neural networks\n\n # Update rule for each parameter\n for l in range(L):\n ### START CODE HERE ### (approx. 2 lines)\n parameters[\"W\" + str(l + 1)] = parameters[\"W\" + str(l + 1)] - learning_rate * grads[\"dW\" + str(l + 1)]\n parameters[\"b\" + str(l + 1)] = parameters[\"b\" + str(l + 1)] - learning_rate * grads[\"db\" + str(l + 1)]\n ### END CODE HERE ###\n \n return parameters", "_____no_output_____" ], [ "parameters, grads, learning_rate = update_parameters_with_gd_test_case()\n\nparameters = update_parameters_with_gd(parameters, grads, learning_rate)\nprint(\"W1 = \" + str(parameters[\"W1\"]))\nprint(\"b1 = \" + str(parameters[\"b1\"]))\nprint(\"W2 = \" + str(parameters[\"W2\"]))\nprint(\"b2 = \" + str(parameters[\"b2\"]))", "W1 = [[ 1.63535156 -0.62320365 -0.53718766]\n [-1.07799357 0.85639907 -2.29470142]]\nb1 = [[ 1.74604067]\n [-0.75184921]]\nW2 = [[ 0.32171798 -0.25467393 1.46902454]\n [-2.05617317 -0.31554548 -0.3756023 ]\n [ 1.1404819 -1.09976462 -0.1612551 ]]\nb2 = [[-0.88020257]\n [ 0.02561572]\n [ 0.57539477]]\n" ] ], [ [ "**Expected Output**:\n\n<table> \n <tr>\n <td > **W1** </td> \n <td > [[ 1.63535156 -0.62320365 -0.53718766]\n [-1.07799357 0.85639907 -2.29470142]] </td> \n </tr> \n \n <tr>\n <td > **b1** </td> \n <td > [[ 1.74604067]\n [-0.75184921]] </td> \n </tr> \n \n <tr>\n <td > **W2** </td> \n <td > [[ 0.32171798 -0.25467393 1.46902454]\n [-2.05617317 -0.31554548 -0.3756023 ]\n [ 1.1404819 -1.09976462 -0.1612551 ]] </td> \n </tr> \n \n <tr>\n <td > **b2** </td> \n <td > [[-0.88020257]\n [ 0.02561572]\n [ 0.57539477]] </td> \n </tr> \n</table>\n", "_____no_output_____" ], [ "A variant of this is Stochastic Gradient Descent (SGD), which is equivalent to mini-batch gradient descent where each mini-batch has just 1 example. The update rule that you have just implemented does not change. What changes is that you would be computing gradients on just one training example at a time, rather than on the whole training set. The code examples below illustrate the difference between stochastic gradient descent and (batch) gradient descent. \n\n- **(Batch) Gradient Descent**:\n\n``` python\nX = data_input\nY = labels\nparameters = initialize_parameters(layers_dims)\nfor i in range(0, num_iterations):\n # Forward propagation\n a, caches = forward_propagation(X, parameters)\n # Compute cost.\n cost = compute_cost(a, Y)\n # Backward propagation.\n grads = backward_propagation(a, caches, parameters)\n # Update parameters.\n parameters = update_parameters(parameters, grads)\n \n```\n\n- **Stochastic Gradient Descent**:\n\n```python\nX = data_input\nY = labels\nparameters = initialize_parameters(layers_dims)\nfor i in range(0, num_iterations):\n for j in range(0, m):\n # Forward propagation\n a, caches = forward_propagation(X[:,j], parameters)\n # Compute cost\n cost = compute_cost(a, Y[:,j])\n # Backward propagation\n grads = backward_propagation(a, caches, parameters)\n # Update parameters.\n parameters = update_parameters(parameters, grads)\n```\n", "_____no_output_____" ], [ "In Stochastic Gradient Descent, you use only 1 training example before updating the gradients. When the training set is large, SGD can be faster. But the parameters will \"oscillate\" toward the minimum rather than converge smoothly. Here is an illustration of this: \n\n<img src=\"images/kiank_sgd.png\" style=\"width:750px;height:250px;\">\n<caption><center> <u> <font color='purple'> **Figure 1** </u><font color='purple'> : **SGD vs GD**<br> \"+\" denotes a minimum of the cost. SGD leads to many oscillations to reach convergence. But each step is a lot faster to compute for SGD than for GD, as it uses only one training example (vs. the whole batch for GD). </center></caption>\n\n**Note** also that implementing SGD requires 3 for-loops in total:\n1. Over the number of iterations\n2. Over the $m$ training examples\n3. Over the layers (to update all parameters, from $(W^{[1]},b^{[1]})$ to $(W^{[L]},b^{[L]})$)\n\nIn practice, you'll often get faster results if you do not use neither the whole training set, nor only one training example, to perform each update. Mini-batch gradient descent uses an intermediate number of examples for each step. With mini-batch gradient descent, you loop over the mini-batches instead of looping over individual training examples.\n\n<img src=\"images/kiank_minibatch.png\" style=\"width:750px;height:250px;\">\n<caption><center> <u> <font color='purple'> **Figure 2** </u>: <font color='purple'> **SGD vs Mini-Batch GD**<br> \"+\" denotes a minimum of the cost. Using mini-batches in your optimization algorithm often leads to faster optimization. </center></caption>\n\n<font color='blue'>\n**What you should remember**:\n- The difference between gradient descent, mini-batch gradient descent and stochastic gradient descent is the number of examples you use to perform one update step.\n- You have to tune a learning rate hyperparameter $\\alpha$.\n- With a well-turned mini-batch size, usually it outperforms either gradient descent or stochastic gradient descent (particularly when the training set is large).", "_____no_output_____" ], [ "## 2 - Mini-Batch Gradient descent\n\nLet's learn how to build mini-batches from the training set (X, Y).\n\nThere are two steps:\n- **Shuffle**: Create a shuffled version of the training set (X, Y) as shown below. Each column of X and Y represents a training example. Note that the random shuffling is done synchronously between X and Y. Such that after the shuffling the $i^{th}$ column of X is the example corresponding to the $i^{th}$ label in Y. The shuffling step ensures that examples will be split randomly into different mini-batches. \n\n<img src=\"images/kiank_shuffle.png\" style=\"width:550px;height:300px;\">\n\n- **Partition**: Partition the shuffled (X, Y) into mini-batches of size `mini_batch_size` (here 64). Note that the number of training examples is not always divisible by `mini_batch_size`. The last mini batch might be smaller, but you don't need to worry about this. When the final mini-batch is smaller than the full `mini_batch_size`, it will look like this: \n\n<img src=\"images/kiank_partition.png\" style=\"width:550px;height:300px;\">\n\n**Exercise**: Implement `random_mini_batches`. We coded the shuffling part for you. To help you with the partitioning step, we give you the following code that selects the indexes for the $1^{st}$ and $2^{nd}$ mini-batches:\n```python\nfirst_mini_batch_X = shuffled_X[:, 0 : mini_batch_size]\nsecond_mini_batch_X = shuffled_X[:, mini_batch_size : 2 * mini_batch_size]\n...\n```\n\nNote that the last mini-batch might end up smaller than `mini_batch_size=64`. Let $\\lfloor s \\rfloor$ represents $s$ rounded down to the nearest integer (this is `math.floor(s)` in Python). If the total number of examples is not a multiple of `mini_batch_size=64` then there will be $\\lfloor \\frac{m}{mini\\_batch\\_size}\\rfloor$ mini-batches with a full 64 examples, and the number of examples in the final mini-batch will be ($m-mini_\\_batch_\\_size \\times \\lfloor \\frac{m}{mini\\_batch\\_size}\\rfloor$). ", "_____no_output_____" ] ], [ [ "# GRADED FUNCTION: random_mini_batches\n\ndef random_mini_batches(X, Y, mini_batch_size = 64, seed = 0):\n \"\"\"\n Creates a list of random minibatches from (X, Y)\n \n Arguments:\n X -- input data, of shape (input size, number of examples)\n Y -- true \"label\" vector (1 for blue dot / 0 for red dot), of shape (1, number of examples)\n mini_batch_size -- size of the mini-batches, integer\n \n Returns:\n mini_batches -- list of synchronous (mini_batch_X, mini_batch_Y)\n \"\"\"\n \n np.random.seed(seed) # To make your \"random\" minibatches the same as ours\n m = X.shape[1] # number of training examples\n mini_batches = []\n \n # Step 1: Shuffle (X, Y)\n permutation = list(np.random.permutation(m))\n shuffled_X = X[:, permutation]\n shuffled_Y = Y[:, permutation].reshape((1,m))\n\n # Step 2: Partition (shuffled_X, shuffled_Y). Minus the end case.\n num_complete_minibatches = math.floor(m/mini_batch_size) # number of mini batches of size mini_batch_size in your partitionning\n for k in range(0, num_complete_minibatches):\n ### START CODE HERE ### (approx. 2 lines)\n mini_batch_X = shuffled_X[:,k * mini_batch_size:(k + 1) * mini_batch_size]\n mini_batch_Y = shuffled_Y[:,k * mini_batch_size:(k + 1) * mini_batch_size]\n ### END CODE HERE ###\n mini_batch = (mini_batch_X, mini_batch_Y)\n mini_batches.append(mini_batch)\n \n # Handling the end case (last mini-batch < mini_batch_size)\n if m % mini_batch_size != 0:\n ### START CODE HERE ### (approx. 2 lines)\n end = m - mini_batch_size * math.floor(m / mini_batch_size)\n mini_batch_X = shuffled_X[:,num_complete_minibatches * mini_batch_size:]\n mini_batch_Y = shuffled_Y[:,num_complete_minibatches * mini_batch_size:]\n ### END CODE HERE ###\n mini_batch = (mini_batch_X, mini_batch_Y)\n mini_batches.append(mini_batch)\n \n return mini_batches", "_____no_output_____" ], [ "X_assess, Y_assess, mini_batch_size = random_mini_batches_test_case()\nmini_batches = random_mini_batches(X_assess, Y_assess, mini_batch_size)\n\nprint(\"shape of the 1st mini_batch_X: \" + str(mini_batches[0][0].shape))\nprint(\"shape of the 2nd mini_batch_X: \" + str(mini_batches[1][0].shape))\nprint(\"shape of the 3rd mini_batch_X: \" + str(mini_batches[2][0].shape))\nprint(\"shape of the 1st mini_batch_Y: \" + str(mini_batches[0][1].shape))\nprint(\"shape of the 2nd mini_batch_Y: \" + str(mini_batches[1][1].shape)) \nprint(\"shape of the 3rd mini_batch_Y: \" + str(mini_batches[2][1].shape))\nprint(\"mini batch sanity check: \" + str(mini_batches[0][0][0][0:3]))", "shape of the 1st mini_batch_X: (12288, 64)\nshape of the 2nd mini_batch_X: (12288, 64)\nshape of the 3rd mini_batch_X: (12288, 20)\nshape of the 1st mini_batch_Y: (1, 64)\nshape of the 2nd mini_batch_Y: (1, 64)\nshape of the 3rd mini_batch_Y: (1, 20)\nmini batch sanity check: [ 0.90085595 -0.7612069 0.2344157 ]\n" ] ], [ [ "**Expected Output**:\n\n<table style=\"width:50%\"> \n <tr>\n <td > **shape of the 1st mini_batch_X** </td> \n <td > (12288, 64) </td> \n </tr> \n \n <tr>\n <td > **shape of the 2nd mini_batch_X** </td> \n <td > (12288, 64) </td> \n </tr> \n \n <tr>\n <td > **shape of the 3rd mini_batch_X** </td> \n <td > (12288, 20) </td> \n </tr>\n <tr>\n <td > **shape of the 1st mini_batch_Y** </td> \n <td > (1, 64) </td> \n </tr> \n <tr>\n <td > **shape of the 2nd mini_batch_Y** </td> \n <td > (1, 64) </td> \n </tr> \n <tr>\n <td > **shape of the 3rd mini_batch_Y** </td> \n <td > (1, 20) </td> \n </tr> \n <tr>\n <td > **mini batch sanity check** </td> \n <td > [ 0.90085595 -0.7612069 0.2344157 ] </td> \n </tr>\n \n</table>", "_____no_output_____" ], [ "<font color='blue'>\n**What you should remember**:\n- Shuffling and Partitioning are the two steps required to build mini-batches\n- Powers of two are often chosen to be the mini-batch size, e.g., 16, 32, 64, 128.", "_____no_output_____" ], [ "## 3 - Momentum\n\nBecause mini-batch gradient descent makes a parameter update after seeing just a subset of examples, the direction of the update has some variance, and so the path taken by mini-batch gradient descent will \"oscillate\" toward convergence. Using momentum can reduce these oscillations. \n\nMomentum takes into account the past gradients to smooth out the update. We will store the 'direction' of the previous gradients in the variable $v$. Formally, this will be the exponentially weighted average of the gradient on previous steps. You can also think of $v$ as the \"velocity\" of a ball rolling downhill, building up speed (and momentum) according to the direction of the gradient/slope of the hill. \n\n<img src=\"images/opt_momentum.png\" style=\"width:400px;height:250px;\">\n<caption><center> <u><font color='purple'>**Figure 3**</u><font color='purple'>: The red arrows shows the direction taken by one step of mini-batch gradient descent with momentum. The blue points show the direction of the gradient (with respect to the current mini-batch) on each step. Rather than just following the gradient, we let the gradient influence $v$ and then take a step in the direction of $v$.<br> <font color='black'> </center>\n\n\n**Exercise**: Initialize the velocity. The velocity, $v$, is a python dictionary that needs to be initialized with arrays of zeros. Its keys are the same as those in the `grads` dictionary, that is:\nfor $l =1,...,L$:\n```python\nv[\"dW\" + str(l+1)] = ... #(numpy array of zeros with the same shape as parameters[\"W\" + str(l+1)])\nv[\"db\" + str(l+1)] = ... #(numpy array of zeros with the same shape as parameters[\"b\" + str(l+1)])\n```\n**Note** that the iterator l starts at 0 in the for loop while the first parameters are v[\"dW1\"] and v[\"db1\"] (that's a \"one\" on the superscript). This is why we are shifting l to l+1 in the `for` loop.", "_____no_output_____" ] ], [ [ "# GRADED FUNCTION: initialize_velocity\n\ndef initialize_velocity(parameters):\n \"\"\"\n Initializes the velocity as a python dictionary with:\n - keys: \"dW1\", \"db1\", ..., \"dWL\", \"dbL\" \n - values: numpy arrays of zeros of the same shape as the corresponding gradients/parameters.\n Arguments:\n parameters -- python dictionary containing your parameters.\n parameters['W' + str(l)] = Wl\n parameters['b' + str(l)] = bl\n \n Returns:\n v -- python dictionary containing the current velocity.\n v['dW' + str(l)] = velocity of dWl\n v['db' + str(l)] = velocity of dbl\n \"\"\"\n \n L = len(parameters) // 2 # number of layers in the neural networks\n v = {}\n \n # Initialize velocity\n for l in range(L):\n ### START CODE HERE ### (approx. 2 lines)\n v[\"dW\" + str(l + 1)] = np.zeros_like(parameters[\"W\" + str(l+1)])\n v[\"db\" + str(l + 1)] = np.zeros_like(parameters[\"b\" + str(l+1)])\n ### END CODE HERE ###\n \n return v", "_____no_output_____" ], [ "parameters = initialize_velocity_test_case()\n\nv = initialize_velocity(parameters)\nprint(\"v[\\\"dW1\\\"] = \" + str(v[\"dW1\"]))\nprint(\"v[\\\"db1\\\"] = \" + str(v[\"db1\"]))\nprint(\"v[\\\"dW2\\\"] = \" + str(v[\"dW2\"]))\nprint(\"v[\\\"db2\\\"] = \" + str(v[\"db2\"]))", "v[\"dW1\"] = [[ 0. 0. 0.]\n [ 0. 0. 0.]]\nv[\"db1\"] = [[ 0.]\n [ 0.]]\nv[\"dW2\"] = [[ 0. 0. 0.]\n [ 0. 0. 0.]\n [ 0. 0. 0.]]\nv[\"db2\"] = [[ 0.]\n [ 0.]\n [ 0.]]\n" ] ], [ [ "**Expected Output**:\n\n<table style=\"width:40%\"> \n <tr>\n <td > **v[\"dW1\"]** </td> \n <td > [[ 0. 0. 0.]\n [ 0. 0. 0.]] </td> \n </tr> \n \n <tr>\n <td > **v[\"db1\"]** </td> \n <td > [[ 0.]\n [ 0.]] </td> \n </tr> \n \n <tr>\n <td > **v[\"dW2\"]** </td> \n <td > [[ 0. 0. 0.]\n [ 0. 0. 0.]\n [ 0. 0. 0.]] </td> \n </tr> \n \n <tr>\n <td > **v[\"db2\"]** </td> \n <td > [[ 0.]\n [ 0.]\n [ 0.]] </td> \n </tr> \n</table>\n", "_____no_output_____" ], [ "**Exercise**: Now, implement the parameters update with momentum. The momentum update rule is, for $l = 1, ..., L$: \n\n$$ \\begin{cases}\nv_{dW^{[l]}} = \\beta v_{dW^{[l]}} + (1 - \\beta) dW^{[l]} \\\\\nW^{[l]} = W^{[l]} - \\alpha v_{dW^{[l]}}\n\\end{cases}\\tag{3}$$\n\n$$\\begin{cases}\nv_{db^{[l]}} = \\beta v_{db^{[l]}} + (1 - \\beta) db^{[l]} \\\\\nb^{[l]} = b^{[l]} - \\alpha v_{db^{[l]}} \n\\end{cases}\\tag{4}$$\n\nwhere L is the number of layers, $\\beta$ is the momentum and $\\alpha$ is the learning rate. All parameters should be stored in the `parameters` dictionary. Note that the iterator `l` starts at 0 in the `for` loop while the first parameters are $W^{[1]}$ and $b^{[1]}$ (that's a \"one\" on the superscript). So you will need to shift `l` to `l+1` when coding.", "_____no_output_____" ] ], [ [ "# GRADED FUNCTION: update_parameters_with_momentum\n\ndef update_parameters_with_momentum(parameters, grads, v, beta, learning_rate):\n \"\"\"\n Update parameters using Momentum\n \n Arguments:\n parameters -- python dictionary containing your parameters:\n parameters['W' + str(l)] = Wl\n parameters['b' + str(l)] = bl\n grads -- python dictionary containing your gradients for each parameters:\n grads['dW' + str(l)] = dWl\n grads['db' + str(l)] = dbl\n v -- python dictionary containing the current velocity:\n v['dW' + str(l)] = ...\n v['db' + str(l)] = ...\n beta -- the momentum hyperparameter, scalar\n learning_rate -- the learning rate, scalar\n \n Returns:\n parameters -- python dictionary containing your updated parameters \n v -- python dictionary containing your updated velocities\n \"\"\"\n\n L = len(parameters) // 2 # number of layers in the neural networks\n \n # Momentum update for each parameter\n for l in range(L):\n \n ### START CODE HERE ### (approx. 4 lines)\n # compute velocities\n v[\"dW\" + str(l + 1)] = beta * v[\"dW\" + str(l + 1)] + (1 - beta) * grads['dW' + str(l + 1)]\n v[\"db\" + str(l + 1)] = beta * v[\"db\" + str(l + 1)] + (1 - beta) * grads['db' + str(l + 1)]\n # update parameters\n parameters[\"W\" + str(l + 1)] = parameters[\"W\" + str(l + 1)] - learning_rate * v[\"dW\" + str(l + 1)]\n parameters[\"b\" + str(l + 1)] = parameters[\"b\" + str(l + 1)] - learning_rate * v[\"db\" + str(l + 1)]\n ### END CODE HERE ###\n \n return parameters, v", "_____no_output_____" ], [ "parameters, grads, v = update_parameters_with_momentum_test_case()\n\nparameters, v = update_parameters_with_momentum(parameters, grads, v, beta = 0.9, learning_rate = 0.01)\nprint(\"W1 = \" + str(parameters[\"W1\"]))\nprint(\"b1 = \" + str(parameters[\"b1\"]))\nprint(\"W2 = \" + str(parameters[\"W2\"]))\nprint(\"b2 = \" + str(parameters[\"b2\"]))\nprint(\"v[\\\"dW1\\\"] = \" + str(v[\"dW1\"]))\nprint(\"v[\\\"db1\\\"] = \" + str(v[\"db1\"]))\nprint(\"v[\\\"dW2\\\"] = \" + str(v[\"dW2\"]))\nprint(\"v[\\\"db2\\\"] = \" + str(v[\"db2\"]))", "W1 = [[ 1.62544598 -0.61290114 -0.52907334]\n [-1.07347112 0.86450677 -2.30085497]]\nb1 = [[ 1.74493465]\n [-0.76027113]]\nW2 = [[ 0.31930698 -0.24990073 1.4627996 ]\n [-2.05974396 -0.32173003 -0.38320915]\n [ 1.13444069 -1.0998786 -0.1713109 ]]\nb2 = [[-0.87809283]\n [ 0.04055394]\n [ 0.58207317]]\nv[\"dW1\"] = [[-0.11006192 0.11447237 0.09015907]\n [ 0.05024943 0.09008559 -0.06837279]]\nv[\"db1\"] = [[-0.01228902]\n [-0.09357694]]\nv[\"dW2\"] = [[-0.02678881 0.05303555 -0.06916608]\n [-0.03967535 -0.06871727 -0.08452056]\n [-0.06712461 -0.00126646 -0.11173103]]\nv[\"db2\"] = [[ 0.02344157]\n [ 0.16598022]\n [ 0.07420442]]\n" ] ], [ [ "**Expected Output**:\n\n<table style=\"width:90%\"> \n <tr>\n <td > **W1** </td> \n <td > [[ 1.62544598 -0.61290114 -0.52907334]\n [-1.07347112 0.86450677 -2.30085497]] </td> \n </tr> \n \n <tr>\n <td > **b1** </td> \n <td > [[ 1.74493465]\n [-0.76027113]] </td> \n </tr> \n \n <tr>\n <td > **W2** </td> \n <td > [[ 0.31930698 -0.24990073 1.4627996 ]\n [-2.05974396 -0.32173003 -0.38320915]\n [ 1.13444069 -1.0998786 -0.1713109 ]] </td> \n </tr> \n \n <tr>\n <td > **b2** </td> \n <td > [[-0.87809283]\n [ 0.04055394]\n [ 0.58207317]] </td> \n </tr> \n\n <tr>\n <td > **v[\"dW1\"]** </td> \n <td > [[-0.11006192 0.11447237 0.09015907]\n [ 0.05024943 0.09008559 -0.06837279]] </td> \n </tr> \n \n <tr>\n <td > **v[\"db1\"]** </td> \n <td > [[-0.01228902]\n [-0.09357694]] </td> \n </tr> \n \n <tr>\n <td > **v[\"dW2\"]** </td> \n <td > [[-0.02678881 0.05303555 -0.06916608]\n [-0.03967535 -0.06871727 -0.08452056]\n [-0.06712461 -0.00126646 -0.11173103]] </td> \n </tr> \n \n <tr>\n <td > **v[\"db2\"]** </td> \n <td > [[ 0.02344157]\n [ 0.16598022]\n [ 0.07420442]]</td> \n </tr> \n</table>\n\n", "_____no_output_____" ], [ "**Note** that:\n- The velocity is initialized with zeros. So the algorithm will take a few iterations to \"build up\" velocity and start to take bigger steps.\n- If $\\beta = 0$, then this just becomes standard gradient descent without momentum. \n\n**How do you choose $\\beta$?**\n\n- The larger the momentum $\\beta$ is, the smoother the update because the more we take the past gradients into account. But if $\\beta$ is too big, it could also smooth out the updates too much. \n- Common values for $\\beta$ range from 0.8 to 0.999. If you don't feel inclined to tune this, $\\beta = 0.9$ is often a reasonable default. \n- Tuning the optimal $\\beta$ for your model might need trying several values to see what works best in term of reducing the value of the cost function $J$. ", "_____no_output_____" ], [ "<font color='blue'>\n**What you should remember**:\n- Momentum takes past gradients into account to smooth out the steps of gradient descent. It can be applied with batch gradient descent, mini-batch gradient descent or stochastic gradient descent.\n- You have to tune a momentum hyperparameter $\\beta$ and a learning rate $\\alpha$.", "_____no_output_____" ], [ "## 4 - Adam\n\nAdam is one of the most effective optimization algorithms for training neural networks. It combines ideas from RMSProp (described in lecture) and Momentum. \n\n**How does Adam work?**\n1. It calculates an exponentially weighted average of past gradients, and stores it in variables $v$ (before bias correction) and $v^{corrected}$ (with bias correction). \n2. It calculates an exponentially weighted average of the squares of the past gradients, and stores it in variables $s$ (before bias correction) and $s^{corrected}$ (with bias correction). \n3. It updates parameters in a direction based on combining information from \"1\" and \"2\".\n\nThe update rule is, for $l = 1, ..., L$: \n\n$$\\begin{cases}\nv_{dW^{[l]}} = \\beta_1 v_{dW^{[l]}} + (1 - \\beta_1) \\frac{\\partial \\mathcal{J} }{ \\partial W^{[l]} } \\\\\nv^{corrected}_{dW^{[l]}} = \\frac{v_{dW^{[l]}}}{1 - (\\beta_1)^t} \\\\\ns_{dW^{[l]}} = \\beta_2 s_{dW^{[l]}} + (1 - \\beta_2) (\\frac{\\partial \\mathcal{J} }{\\partial W^{[l]} })^2 \\\\\ns^{corrected}_{dW^{[l]}} = \\frac{s_{dW^{[l]}}}{1 - (\\beta_1)^t} \\\\\nW^{[l]} = W^{[l]} - \\alpha \\frac{v^{corrected}_{dW^{[l]}}}{\\sqrt{s^{corrected}_{dW^{[l]}}} + \\varepsilon}\n\\end{cases}$$\nwhere:\n- t counts the number of steps taken of Adam \n- L is the number of layers\n- $\\beta_1$ and $\\beta_2$ are hyperparameters that control the two exponentially weighted averages. \n- $\\alpha$ is the learning rate\n- $\\varepsilon$ is a very small number to avoid dividing by zero\n\nAs usual, we will store all parameters in the `parameters` dictionary ", "_____no_output_____" ], [ "**Exercise**: Initialize the Adam variables $v, s$ which keep track of the past information.\n\n**Instruction**: The variables $v, s$ are python dictionaries that need to be initialized with arrays of zeros. Their keys are the same as for `grads`, that is:\nfor $l = 1, ..., L$:\n```python\nv[\"dW\" + str(l+1)] = ... #(numpy array of zeros with the same shape as parameters[\"W\" + str(l+1)])\nv[\"db\" + str(l+1)] = ... #(numpy array of zeros with the same shape as parameters[\"b\" + str(l+1)])\ns[\"dW\" + str(l+1)] = ... #(numpy array of zeros with the same shape as parameters[\"W\" + str(l+1)])\ns[\"db\" + str(l+1)] = ... #(numpy array of zeros with the same shape as parameters[\"b\" + str(l+1)])\n\n```", "_____no_output_____" ] ], [ [ "# GRADED FUNCTION: initialize_adam\n\ndef initialize_adam(parameters) :\n \"\"\"\n Initializes v and s as two python dictionaries with:\n - keys: \"dW1\", \"db1\", ..., \"dWL\", \"dbL\" \n - values: numpy arrays of zeros of the same shape as the corresponding gradients/parameters.\n \n Arguments:\n parameters -- python dictionary containing your parameters.\n parameters[\"W\" + str(l)] = Wl\n parameters[\"b\" + str(l)] = bl\n \n Returns: \n v -- python dictionary that will contain the exponentially weighted average of the gradient.\n v[\"dW\" + str(l)] = ...\n v[\"db\" + str(l)] = ...\n s -- python dictionary that will contain the exponentially weighted average of the squared gradient.\n s[\"dW\" + str(l)] = ...\n s[\"db\" + str(l)] = ...\n\n \"\"\"\n \n L = len(parameters) // 2 # number of layers in the neural networks\n v = {}\n s = {}\n \n # Initialize v, s. Input: \"parameters\". Outputs: \"v, s\".\n for l in range(L):\n ### START CODE HERE ### (approx. 4 lines)\n v[\"dW\" + str(l + 1)] = np.zeros_like(parameters[\"W\" + str(l + 1)])\n v[\"db\" + str(l + 1)] = np.zeros_like(parameters[\"b\" + str(l + 1)])\n\n s[\"dW\" + str(l+1)] = np.zeros_like(parameters[\"W\" + str(l + 1)])\n s[\"db\" + str(l+1)] = np.zeros_like(parameters[\"b\" + str(l + 1)])\n ### END CODE HERE ###\n \n return v, s", "_____no_output_____" ], [ "parameters = initialize_adam_test_case()\n\nv, s = initialize_adam(parameters)\nprint(\"v[\\\"dW1\\\"] = \" + str(v[\"dW1\"]))\nprint(\"v[\\\"db1\\\"] = \" + str(v[\"db1\"]))\nprint(\"v[\\\"dW2\\\"] = \" + str(v[\"dW2\"]))\nprint(\"v[\\\"db2\\\"] = \" + str(v[\"db2\"]))\nprint(\"s[\\\"dW1\\\"] = \" + str(s[\"dW1\"]))\nprint(\"s[\\\"db1\\\"] = \" + str(s[\"db1\"]))\nprint(\"s[\\\"dW2\\\"] = \" + str(s[\"dW2\"]))\nprint(\"s[\\\"db2\\\"] = \" + str(s[\"db2\"]))", "v[\"dW1\"] = [[ 0. 0. 0.]\n [ 0. 0. 0.]]\nv[\"db1\"] = [[ 0.]\n [ 0.]]\nv[\"dW2\"] = [[ 0. 0. 0.]\n [ 0. 0. 0.]\n [ 0. 0. 0.]]\nv[\"db2\"] = [[ 0.]\n [ 0.]\n [ 0.]]\ns[\"dW1\"] = [[ 0. 0. 0.]\n [ 0. 0. 0.]]\ns[\"db1\"] = [[ 0.]\n [ 0.]]\ns[\"dW2\"] = [[ 0. 0. 0.]\n [ 0. 0. 0.]\n [ 0. 0. 0.]]\ns[\"db2\"] = [[ 0.]\n [ 0.]\n [ 0.]]\n" ] ], [ [ "**Expected Output**:\n\n<table style=\"width:40%\"> \n <tr>\n <td > **v[\"dW1\"]** </td> \n <td > [[ 0. 0. 0.]\n [ 0. 0. 0.]] </td> \n </tr> \n \n <tr>\n <td > **v[\"db1\"]** </td> \n <td > [[ 0.]\n [ 0.]] </td> \n </tr> \n \n <tr>\n <td > **v[\"dW2\"]** </td> \n <td > [[ 0. 0. 0.]\n [ 0. 0. 0.]\n [ 0. 0. 0.]] </td> \n </tr> \n \n <tr>\n <td > **v[\"db2\"]** </td> \n <td > [[ 0.]\n [ 0.]\n [ 0.]] </td> \n </tr> \n <tr>\n <td > **s[\"dW1\"]** </td> \n <td > [[ 0. 0. 0.]\n [ 0. 0. 0.]] </td> \n </tr> \n \n <tr>\n <td > **s[\"db1\"]** </td> \n <td > [[ 0.]\n [ 0.]] </td> \n </tr> \n \n <tr>\n <td > **s[\"dW2\"]** </td> \n <td > [[ 0. 0. 0.]\n [ 0. 0. 0.]\n [ 0. 0. 0.]] </td> \n </tr> \n \n <tr>\n <td > **s[\"db2\"]** </td> \n <td > [[ 0.]\n [ 0.]\n [ 0.]] </td> \n </tr>\n\n</table>\n", "_____no_output_____" ], [ "**Exercise**: Now, implement the parameters update with Adam. Recall the general update rule is, for $l = 1, ..., L$: \n\n$$\\begin{cases}\nv_{W^{[l]}} = \\beta_1 v_{W^{[l]}} + (1 - \\beta_1) \\frac{\\partial J }{ \\partial W^{[l]} } \\\\\nv^{corrected}_{W^{[l]}} = \\frac{v_{W^{[l]}}}{1 - (\\beta_1)^t} \\\\\ns_{W^{[l]}} = \\beta_2 s_{W^{[l]}} + (1 - \\beta_2) (\\frac{\\partial J }{\\partial W^{[l]} })^2 \\\\\ns^{corrected}_{W^{[l]}} = \\frac{s_{W^{[l]}}}{1 - (\\beta_2)^t} \\\\\nW^{[l]} = W^{[l]} - \\alpha \\frac{v^{corrected}_{W^{[l]}}}{\\sqrt{s^{corrected}_{W^{[l]}}}+\\varepsilon}\n\\end{cases}$$\n\n\n**Note** that the iterator `l` starts at 0 in the `for` loop while the first parameters are $W^{[1]}$ and $b^{[1]}$. You need to shift `l` to `l+1` when coding.", "_____no_output_____" ] ], [ [ "# GRADED FUNCTION: update_parameters_with_adam\n\ndef update_parameters_with_adam(parameters, grads, v, s, t, learning_rate=0.01,\n beta1=0.9, beta2=0.999, epsilon=1e-8):\n \"\"\"\n Update parameters using Adam\n \n Arguments:\n parameters -- python dictionary containing your parameters:\n parameters['W' + str(l)] = Wl\n parameters['b' + str(l)] = bl\n grads -- python dictionary containing your gradients for each parameters:\n grads['dW' + str(l)] = dWl\n grads['db' + str(l)] = dbl\n v -- Adam variable, moving average of the first gradient, python dictionary\n s -- Adam variable, moving average of the squared gradient, python dictionary\n learning_rate -- the learning rate, scalar.\n beta1 -- Exponential decay hyperparameter for the first moment estimates \n beta2 -- Exponential decay hyperparameter for the second moment estimates \n epsilon -- hyperparameter preventing division by zero in Adam updates\n\n Returns:\n parameters -- python dictionary containing your updated parameters \n v -- Adam variable, moving average of the first gradient, python dictionary\n s -- Adam variable, moving average of the squared gradient, python dictionary\n \"\"\"\n \n L = len(parameters) // 2 # number of layers in the neural networks\n v_corrected = {} # Initializing first moment estimate, python dictionary\n s_corrected = {} # Initializing second moment estimate, python dictionary\n \n # Perform Adam update on all parameters\n for l in range(L):\n # Moving average of the gradients. Inputs: \"v, grads, beta1\". Output: \"v\".\n ### START CODE HERE ### (approx. 2 lines)\n v[\"dW\" + str(l + 1)] = beta1 * v[\"dW\" + str(l + 1)] + (1 - beta1) * grads['dW' + str(l + 1)]\n v[\"db\" + str(l + 1)] = beta1 * v[\"db\" + str(l + 1)] + (1 - beta1) * grads['db' + str(l + 1)]\n ### END CODE HERE ###\n\n # Compute bias-corrected first moment estimate. Inputs: \"v, beta1, t\". Output: \"v_corrected\".\n ### START CODE HERE ### (approx. 2 lines)\n v_corrected[\"dW\" + str(l + 1)] = v[\"dW\" + str(l + 1)] / (1 - np.power(beta1, t))\n v_corrected[\"db\" + str(l + 1)] = v[\"db\" + str(l + 1)] / (1 - np.power(beta1, t))\n ### END CODE HERE ###\n\n # Moving average of the squared gradients. Inputs: \"s, grads, beta2\". Output: \"s\".\n ### START CODE HERE ### (approx. 2 lines)\n s[\"dW\" + str(l + 1)] = beta2 * s[\"dW\" + str(l + 1)] + (1 - beta2) * np.power(grads['dW' + str(l + 1)], 2)\n s[\"db\" + str(l + 1)] = beta2 * s[\"db\" + str(l + 1)] + (1 - beta2) * np.power(grads['db' + str(l + 1)], 2)\n ### END CODE HERE ###\n\n # Compute bias-corrected second raw moment estimate. Inputs: \"s, beta2, t\". Output: \"s_corrected\".\n ### START CODE HERE ### (approx. 2 lines)\n s_corrected[\"dW\" + str(l + 1)] = s[\"dW\" + str(l + 1)] / (1 - np.power(beta2, t))\n s_corrected[\"db\" + str(l + 1)] = s[\"db\" + str(l + 1)] / (1 - np.power(beta2, t))\n ### END CODE HERE ###\n\n # Update parameters. Inputs: \"parameters, learning_rate, v_corrected, s_corrected, epsilon\". Output: \"parameters\".\n ### START CODE HERE ### (approx. 2 lines)\n parameters[\"W\" + str(l + 1)] = parameters[\"W\" + str(l + 1)] - learning_rate * v_corrected[\"dW\" + str(l + 1)] / np.sqrt(s_corrected[\"dW\" + str(l + 1)] + epsilon)\n parameters[\"b\" + str(l + 1)] = parameters[\"b\" + str(l + 1)] - learning_rate * v_corrected[\"db\" + str(l + 1)] / np.sqrt(s_corrected[\"db\" + str(l + 1)] + epsilon)\n ### END CODE HERE ###\n\n return parameters, v, s", "_____no_output_____" ], [ "parameters, grads, v, s = update_parameters_with_adam_test_case()\nparameters, v, s = update_parameters_with_adam(parameters, grads, v, s, t = 2)\n\nprint(\"W1 = \" + str(parameters[\"W1\"]))\nprint(\"b1 = \" + str(parameters[\"b1\"]))\nprint(\"W2 = \" + str(parameters[\"W2\"]))\nprint(\"b2 = \" + str(parameters[\"b2\"]))\nprint(\"v[\\\"dW1\\\"] = \" + str(v[\"dW1\"]))\nprint(\"v[\\\"db1\\\"] = \" + str(v[\"db1\"]))\nprint(\"v[\\\"dW2\\\"] = \" + str(v[\"dW2\"]))\nprint(\"v[\\\"db2\\\"] = \" + str(v[\"db2\"]))\nprint(\"s[\\\"dW1\\\"] = \" + str(s[\"dW1\"]))\nprint(\"s[\\\"db1\\\"] = \" + str(s[\"db1\"]))\nprint(\"s[\\\"dW2\\\"] = \" + str(s[\"dW2\"]))\nprint(\"s[\\\"db2\\\"] = \" + str(s[\"db2\"]))", "W1 = [[ 1.79078034 -0.77819144 -0.69460639]\n [-1.23940099 0.69897299 -2.13510481]]\nb1 = [[ 1.91119235]\n [-0.59477218]]\nW2 = [[ 0.48546317 -0.41580308 1.62854186]\n [-1.89371033 -0.1559833 -0.21761985]\n [ 1.30020326 -0.93841334 -0.00599321]]\nb2 = [[-1.04427894]\n [-0.12422162]\n [ 0.41638106]]\nv[\"dW1\"] = [[-0.11006192 0.11447237 0.09015907]\n [ 0.05024943 0.09008559 -0.06837279]]\nv[\"db1\"] = [[-0.01228902]\n [-0.09357694]]\nv[\"dW2\"] = [[-0.02678881 0.05303555 -0.06916608]\n [-0.03967535 -0.06871727 -0.08452056]\n [-0.06712461 -0.00126646 -0.11173103]]\nv[\"db2\"] = [[ 0.02344157]\n [ 0.16598022]\n [ 0.07420442]]\ns[\"dW1\"] = [[ 0.00121136 0.00131039 0.00081287]\n [ 0.0002525 0.00081154 0.00046748]]\ns[\"db1\"] = [[ 1.51020075e-05]\n [ 8.75664434e-04]]\ns[\"dW2\"] = [[ 7.17640232e-05 2.81276921e-04 4.78394595e-04]\n [ 1.57413361e-04 4.72206320e-04 7.14372576e-04]\n [ 4.50571368e-04 1.60392066e-07 1.24838242e-03]]\ns[\"db2\"] = [[ 5.49507194e-05]\n [ 2.75494327e-03]\n [ 5.50629536e-04]]\n" ] ], [ [ "**Expected Output**:\n\n<table> \n <tr>\n <td > **W1** </td> \n <td > [[ 1.63178673 -0.61919778 -0.53561312]\n [-1.08040999 0.85796626 -2.29409733]] </td> \n </tr> \n \n <tr>\n <td > **b1** </td> \n <td > [[ 1.75225313]\n [-0.75376553]] </td> \n </tr> \n \n <tr>\n <td > **W2** </td> \n <td > [[ 0.32648046 -0.25681174 1.46954931]\n [-2.05269934 -0.31497584 -0.37661299]\n [ 1.14121081 -1.09245036 -0.16498684]] </td> \n </tr> \n \n <tr>\n <td > **b2** </td> \n <td > [[-0.88529978]\n [ 0.03477238]\n [ 0.57537385]] </td> \n </tr> \n <tr>\n <td > **v[\"dW1\"]** </td> \n <td > [[-0.11006192 0.11447237 0.09015907]\n [ 0.05024943 0.09008559 -0.06837279]] </td> \n </tr> \n \n <tr>\n <td > **v[\"db1\"]** </td> \n <td > [[-0.01228902]\n [-0.09357694]] </td> \n </tr> \n \n <tr>\n <td > **v[\"dW2\"]** </td> \n <td > [[-0.02678881 0.05303555 -0.06916608]\n [-0.03967535 -0.06871727 -0.08452056]\n [-0.06712461 -0.00126646 -0.11173103]] </td> \n </tr> \n \n <tr>\n <td > **v[\"db2\"]** </td> \n <td > [[ 0.02344157]\n [ 0.16598022]\n [ 0.07420442]] </td> \n </tr> \n <tr>\n <td > **s[\"dW1\"]** </td> \n <td > [[ 0.00121136 0.00131039 0.00081287]\n [ 0.0002525 0.00081154 0.00046748]] </td> \n </tr> \n \n <tr>\n <td > **s[\"db1\"]** </td> \n <td > [[ 1.51020075e-05]\n [ 8.75664434e-04]] </td> \n </tr> \n \n <tr>\n <td > **s[\"dW2\"]** </td> \n <td > [[ 7.17640232e-05 2.81276921e-04 4.78394595e-04]\n [ 1.57413361e-04 4.72206320e-04 7.14372576e-04]\n [ 4.50571368e-04 1.60392066e-07 1.24838242e-03]] </td> \n </tr> \n \n <tr>\n <td > **s[\"db2\"]** </td> \n <td > [[ 5.49507194e-05]\n [ 2.75494327e-03]\n [ 5.50629536e-04]] </td> \n </tr>\n</table>\n", "_____no_output_____" ], [ "You now have three working optimization algorithms (mini-batch gradient descent, Momentum, Adam). Let's implement a model with each of these optimizers and observe the difference.", "_____no_output_____" ], [ "## 5 - Model with different optimization algorithms\n\nLets use the following \"moons\" dataset to test the different optimization methods. (The dataset is named \"moons\" because the data from each of the two classes looks a bit like a crescent-shaped moon.) ", "_____no_output_____" ] ], [ [ "train_X, train_Y = load_dataset()", "_____no_output_____" ] ], [ [ "We have already implemented a 3-layer neural network. You will train it with: \n- Mini-batch **Gradient Descent**: it will call your function:\n - `update_parameters_with_gd()`\n- Mini-batch **Momentum**: it will call your functions:\n - `initialize_velocity()` and `update_parameters_with_momentum()`\n- Mini-batch **Adam**: it will call your functions:\n - `initialize_adam()` and `update_parameters_with_adam()`", "_____no_output_____" ] ], [ [ "def model(X, Y, layers_dims, optimizer, learning_rate=0.0007, mini_batch_size=64, beta=0.9,\n beta1=0.9, beta2=0.999, epsilon=1e-8, num_epochs=10000, print_cost=True):\n \"\"\"\n 3-layer neural network model which can be run in different optimizer modes.\n \n Arguments:\n X -- input data, of shape (2, number of examples)\n Y -- true \"label\" vector (1 for blue dot / 0 for red dot), of shape (1, number of examples)\n layers_dims -- python list, containing the size of each layer\n learning_rate -- the learning rate, scalar.\n mini_batch_size -- the size of a mini batch\n beta -- Momentum hyperparameter\n beta1 -- Exponential decay hyperparameter for the past gradients estimates \n beta2 -- Exponential decay hyperparameter for the past squared gradients estimates \n epsilon -- hyperparameter preventing division by zero in Adam updates\n num_epochs -- number of epochs\n print_cost -- True to print the cost every 1000 epochs\n\n Returns:\n parameters -- python dictionary containing your updated parameters \n \"\"\"\n\n L = len(layers_dims) # number of layers in the neural networks\n costs = [] # to keep track of the cost\n t = 0 # initializing the counter required for Adam update\n seed = 10 # For grading purposes, so that your \"random\" minibatches are the same as ours\n \n # Initialize parameters\n parameters = initialize_parameters(layers_dims)\n\n # Initialize the optimizer\n if optimizer == \"gd\":\n pass # no initialization required for gradient descent\n elif optimizer == \"momentum\":\n v = initialize_velocity(parameters)\n elif optimizer == \"adam\":\n v, s = initialize_adam(parameters)\n \n # Optimization loop\n for i in range(num_epochs):\n \n # Define the random minibatches. We increment the seed to reshuffle differently the dataset after each epoch\n seed = seed + 1\n minibatches = random_mini_batches(X, Y, mini_batch_size, seed)\n\n for minibatch in minibatches:\n\n # Select a minibatch\n (minibatch_X, minibatch_Y) = minibatch\n\n # Forward propagation\n a3, caches = forward_propagation(minibatch_X, parameters)\n\n # Compute cost\n cost = compute_cost(a3, minibatch_Y)\n\n # Backward propagation\n grads = backward_propagation(minibatch_X, minibatch_Y, caches)\n\n # Update parameters\n if optimizer == \"gd\":\n parameters = update_parameters_with_gd(parameters, grads, learning_rate)\n elif optimizer == \"momentum\":\n parameters, v = update_parameters_with_momentum(parameters, grads, v, beta, learning_rate)\n elif optimizer == \"adam\":\n t = t + 1 # Adam counter\n parameters, v, s = update_parameters_with_adam(parameters, grads, v, s,\n t, learning_rate, beta1, beta2, epsilon)\n \n # Print the cost every 1000 epoch\n if print_cost and i % 1000 == 0:\n print(\"Cost after epoch %i: %f\" % (i, cost))\n if print_cost and i % 100 == 0:\n costs.append(cost)\n \n # plot the cost\n plt.plot(costs)\n plt.ylabel('cost')\n plt.xlabel('epochs (per 100)')\n plt.title(\"Learning rate = \" + str(learning_rate))\n plt.show()\n\n return parameters", "_____no_output_____" ] ], [ [ "You will now run this 3 layer neural network with each of the 3 optimization methods.\n\n### 5.1 - Mini-batch Gradient descent\n\nRun the following code to see how the model does with mini-batch gradient descent.", "_____no_output_____" ] ], [ [ "# train 3-layer model\nlayers_dims = [train_X.shape[0], 5, 2, 1]\nparameters = model(train_X, train_Y, layers_dims, optimizer=\"gd\")\n\n# Predict\npredictions = predict(train_X, train_Y, parameters)\n\n# Plot decision boundary\nplt.title(\"Model with Gradient Descent optimization\")\naxes = plt.gca()\naxes.set_xlim([-1.5, 2.5])\naxes.set_ylim([-1, 1.5])\nplot_decision_boundary(lambda x: predict_dec(parameters, x.T), train_X, train_Y)", "Cost after epoch 0: 0.690736\nCost after epoch 1000: 0.685273\nCost after epoch 2000: 0.647072\nCost after epoch 3000: 0.619525\nCost after epoch 4000: 0.576584\nCost after epoch 5000: 0.607243\nCost after epoch 6000: 0.529403\nCost after epoch 7000: 0.460768\nCost after epoch 8000: 0.465586\nCost after epoch 9000: 0.464518\n" ] ], [ [ "### 5.2 - Mini-batch gradient descent with momentum\n\nRun the following code to see how the model does with momentum. Because this example is relatively simple, the gains from using momemtum are small; but for more complex problems you might see bigger gains.", "_____no_output_____" ] ], [ [ "# train 3-layer model\nlayers_dims = [train_X.shape[0], 5, 2, 1]\nparameters = model(train_X, train_Y, layers_dims, beta=0.9, optimizer=\"momentum\")\n\n# Predict\npredictions = predict(train_X, train_Y, parameters)\n\n# Plot decision boundary\nplt.title(\"Model with Momentum optimization\")\naxes = plt.gca()\naxes.set_xlim([-1.5, 2.5])\naxes.set_ylim([-1, 1.5])\nplot_decision_boundary(lambda x: predict_dec(parameters, x.T), train_X, train_Y)", "Cost after epoch 0: 0.690741\nCost after epoch 1000: 0.685341\nCost after epoch 2000: 0.647145\nCost after epoch 3000: 0.619594\nCost after epoch 4000: 0.576665\nCost after epoch 5000: 0.607324\nCost after epoch 6000: 0.529476\nCost after epoch 7000: 0.460936\nCost after epoch 8000: 0.465780\nCost after epoch 9000: 0.464740\n" ] ], [ [ "### 5.3 - Mini-batch with Adam mode\n\nRun the following code to see how the model does with Adam.", "_____no_output_____" ] ], [ [ "# train 3-layer model\nlayers_dims = [train_X.shape[0], 5, 2, 1]\nparameters = model(train_X, train_Y, layers_dims, optimizer=\"adam\")\n\n# Predict\npredictions = predict(train_X, train_Y, parameters)\n\n# Plot decision boundary\nplt.title(\"Model with Adam optimization\")\naxes = plt.gca()\naxes.set_xlim([-1.5, 2.5])\naxes.set_ylim([-1, 1.5])\nplot_decision_boundary(lambda x: predict_dec(parameters, x.T), train_X, train_Y)", "Cost after epoch 0: 0.687550\nCost after epoch 1000: 0.173593\nCost after epoch 2000: 0.150145\nCost after epoch 3000: 0.072939\nCost after epoch 4000: 0.125896\nCost after epoch 5000: 0.104185\nCost after epoch 6000: 0.116069\nCost after epoch 7000: 0.031774\nCost after epoch 8000: 0.112908\nCost after epoch 9000: 0.197732\n" ] ], [ [ "### 5.4 - Summary\n\n<table> \n <tr>\n <td>\n **optimization method**\n </td>\n <td>\n **accuracy**\n </td>\n <td>\n **cost shape**\n </td>\n\n </tr>\n <td>\n Gradient descent\n </td>\n <td>\n 79.7%\n </td>\n <td>\n oscillations\n </td>\n <tr>\n <td>\n Momentum\n </td>\n <td>\n 79.7%\n </td>\n <td>\n oscillations\n </td>\n </tr>\n <tr>\n <td>\n Adam\n </td>\n <td>\n 94%\n </td>\n <td>\n smoother\n </td>\n </tr>\n</table> \n\nMomentum usually helps, but given the small learning rate and the simplistic dataset, its impact is almost negligeable. Also, the huge oscillations you see in the cost come from the fact that some minibatches are more difficult thans others for the optimization algorithm.\n\nAdam on the other hand, clearly outperforms mini-batch gradient descent and Momentum. If you run the model for more epochs on this simple dataset, all three methods will lead to very good results. However, you've seen that Adam converges a lot faster.\n\nSome advantages of Adam include:\n- Relatively low memory requirements (though higher than gradient descent and gradient descent with momentum) \n- Usually works well even with little tuning of hyperparameters (except $\\alpha$)", "_____no_output_____" ], [ "**References**:\n\n- Adam paper: https://arxiv.org/pdf/1412.6980.pdf", "_____no_output_____" ] ] ]

[ "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown" ]

[ [ "markdown" ], [ "code" ], [ "markdown" ], [ "code", "code" ], [ "markdown", "markdown", "markdown", "markdown" ], [ "code", "code" ], [ "markdown", "markdown", "markdown" ], [ "code", "code" ], [ "markdown", "markdown" ], [ "code", "code" ], [ "markdown", "markdown", "markdown", "markdown", "markdown" ], [ "code", "code" ], [ "markdown", "markdown" ], [ "code", "code" ], [ "markdown", "markdown", "markdown" ], [ "code" ], [ "markdown" ], [ "code" ], [ "markdown" ], [ "code" ], [ "markdown" ], [ "code" ], [ "markdown" ], [ "code" ], [ "markdown", "markdown" ] ]

[ "MIT" ]

[ "MIT" ]

[ "MIT" ]

[ [ [ "# Python Challenge 3\n\nchallenge URL: http://www.pythonchallenge.com/pc/def/equality.html", "_____no_output_____" ] ], [ [ "import urllib.request\nimport re\n\nhtml = urllib.request.urlopen(\"http://www.pythonchallenge.com/pc/def/equality.html\").read().decode()\ndata = re.findall(\"<!--(.*?)-->\", html, re.DOTALL)[-1]\ndata", "_____no_output_____" ], [ "# ***AAAbCCC***\n# [A-Z] capital letters\n# [a-z] small caps\n# [^] not\n# + multiple\n# {3} 3 of something\n\n\"\".join(re.findall( \"[^A-Z]+[A-Z]{3}([a-z])[A-Z]{3}[^A-Z]+\", data))", "_____no_output_____" ] ], [ [ "# Advanced\n\nhttps://www.geeksforgeeks.org/inplace-rotate-square-matrix-by-90-degrees/", "_____no_output_____" ] ] ]

[ "markdown", "code", "markdown" ]

[ [ "markdown" ], [ "code", "code" ], [ "markdown" ] ]

[ "MIT" ]

[ "MIT" ]

[ "MIT" ]

[ [ [ "# Creating a Sentiment Analysis Web App\n## Using PyTorch and SageMaker\n\n_Deep Learning Nanodegree Program | Deployment_\n\n---\n\nNow that we have a basic understanding of how SageMaker works we will try to use it to construct a complete project from end to end. Our goal will be to have a simple web page which a user can use to enter a movie review. The web page will then send the review off to our deployed model which will predict the sentiment of the entered review.\n\n## Instructions\n\nSome template code has already been provided for you, and you will need to implement additional functionality to successfully complete this notebook. You will not need to modify the included code beyond what is requested. Sections that begin with '**TODO**' in the header indicate that you need to complete or implement some portion within them. Instructions will be provided for each section and the specifics of the implementation are marked in the code block with a `# TODO: ...` comment. Please be sure to read the instructions carefully!\n\nIn addition to implementing code, there will be questions for you to answer which relate to the task and your implementation. Each section where you will answer a question is preceded by a '**Question:**' header. Carefully read each question and provide your answer below the '**Answer:**' header by editing the Markdown cell.\n\n> **Note**: Code and Markdown cells can be executed using the **Shift+Enter** keyboard shortcut. In addition, a cell can be edited by typically clicking it (double-click for Markdown cells) or by pressing **Enter** while it is highlighted.\n\n## General Outline\n\nRecall the general outline for SageMaker projects using a notebook instance.\n\n1. Download or otherwise retrieve the data.\n2. Process / Prepare the data.\n3. Upload the processed data to S3.\n4. Train a chosen model.\n5. Test the trained model (typically using a batch transform job).\n6. Deploy the trained model.\n7. Use the deployed model.\n\nFor this project, you will be following the steps in the general outline with some modifications. \n\nFirst, you will not be testing the model in its own step. You will still be testing the model, however, you will do it by deploying your model and then using the deployed model by sending the test data to it. One of the reasons for doing this is so that you can make sure that your deployed model is working correctly before moving forward.\n\nIn addition, you will deploy and use your trained model a second time. In the second iteration you will customize the way that your trained model is deployed by including some of your own code. In addition, your newly deployed model will be used in the sentiment analysis web app.", "_____no_output_____" ] ], [ [ "# Make sure that we use SageMaker 1.x\n!pip install sagemaker==1.72.0", "Requirement already satisfied: sagemaker==1.72.0 in c:\\users\\katherine\\anaconda3\\lib\\site-packages (1.72.0)\nRequirement already satisfied: importlib-metadata>=1.4.0 in c:\\users\\katherine\\anaconda3\\lib\\site-packages (from sagemaker==1.72.0) (3.10.0)\nRequirement already satisfied: numpy>=1.9.0 in c:\\users\\katherine\\anaconda3\\lib\\site-packages (from sagemaker==1.72.0) (1.20.1)\nRequirement already satisfied: scipy>=0.19.0 in c:\\users\\katherine\\anaconda3\\lib\\site-packages (from sagemaker==1.72.0) (1.6.2)\nRequirement already satisfied: protobuf3-to-dict>=0.1.5 in c:\\users\\katherine\\anaconda3\\lib\\site-packages (from sagemaker==1.72.0) (0.1.5)\nRequirement already satisfied: boto3>=1.14.12 in c:\\users\\katherine\\anaconda3\\lib\\site-packages (from sagemaker==1.72.0) (1.17.97)\nRequirement already satisfied: smdebug-rulesconfig==0.1.4 in c:\\users\\katherine\\anaconda3\\lib\\site-packages (from sagemaker==1.72.0) (0.1.4)\nRequirement already satisfied: packaging>=20.0 in c:\\users\\katherine\\anaconda3\\lib\\site-packages (from sagemaker==1.72.0) (20.9)\nRequirement already satisfied: protobuf>=3.1 in c:\\users\\katherine\\anaconda3\\lib\\site-packages (from sagemaker==1.72.0) (3.17.3)\nRequirement already satisfied: s3transfer<0.5.0,>=0.4.0 in c:\\users\\katherine\\anaconda3\\lib\\site-packages (from boto3>=1.14.12->sagemaker==1.72.0) (0.4.2)\nRequirement already satisfied: botocore<1.21.0,>=1.20.97 in c:\\users\\katherine\\anaconda3\\lib\\site-packages (from boto3>=1.14.12->sagemaker==1.72.0) (1.20.97)\nRequirement already satisfied: jmespath<1.0.0,>=0.7.1 in c:\\users\\katherine\\anaconda3\\lib\\site-packages (from boto3>=1.14.12->sagemaker==1.72.0) (0.10.0)\nRequirement already satisfied: python-dateutil<3.0.0,>=2.1 in c:\\users\\katherine\\anaconda3\\lib\\site-packages (from botocore<1.21.0,>=1.20.97->boto3>=1.14.12->sagemaker==1.72.0) (2.8.1)\nRequirement already satisfied: urllib3<1.27,>=1.25.4 in c:\\users\\katherine\\anaconda3\\lib\\site-packages (from botocore<1.21.0,>=1.20.97->boto3>=1.14.12->sagemaker==1.72.0) (1.26.4)\nRequirement already satisfied: zipp>=0.5 in c:\\users\\katherine\\anaconda3\\lib\\site-packages (from importlib-metadata>=1.4.0->sagemaker==1.72.0) (3.4.1)\nRequirement already satisfied: pyparsing>=2.0.2 in c:\\users\\katherine\\anaconda3\\lib\\site-packages (from packaging>=20.0->sagemaker==1.72.0) (2.4.7)\nRequirement already satisfied: six>=1.9 in c:\\users\\katherine\\anaconda3\\lib\\site-packages (from protobuf>=3.1->sagemaker==1.72.0) (1.15.0)\n" ] ], [ [ "## Step 1: Downloading the data\n\nAs in the XGBoost in SageMaker notebook, we will be using the [IMDb dataset](http://ai.stanford.edu/~amaas/data/sentiment/)\n\n> Maas, Andrew L., et al. [Learning Word Vectors for Sentiment Analysis](http://ai.stanford.edu/~amaas/data/sentiment/). In _Proceedings of the 49th Annual Meeting of the Association for Computational Linguistics: Human Language Technologies_. Association for Computational Linguistics, 2011.", "_____no_output_____" ] ], [ [ "#%mkdir .\\data\n#jupyter notebook --notebook-dir=/Users/Katherine/UdacityMLE/sagemaker-deployment-master/data\n%cd C:/Users/Katherine/UdacityMLE/sagemaker-deployment-master/data", "C:\\Users\\Katherine\\UdacityMLE\\sagemaker-deployment-master\\data\n" ], [ "#!wget -O ./data/aclImdb_v1.tar.gz http://ai.stanford.edu/~amaas/data/sentiment/aclImdb_v1.tar.gz\n!tar -zxf ./data/aclImdb_v1.tar.gz -C ./data", "_____no_output_____" ] ], [ [ "## Step 2: Preparing and Processing the data\n\nAlso, as in the XGBoost notebook, we will be doing some initial data processing. The first few steps are the same as in the XGBoost example. To begin with, we will read in each of the reviews and combine them into a single input structure. Then, we will split the dataset into a training set and a testing set.", "_____no_output_____" ] ], [ [ "import os\nimport glob\n\ndef read_imdb_data(data_dir='./aclImdb'):\n data = {}\n labels = {}\n \n for data_type in ['train', 'test']:\n data[data_type] = {}\n labels[data_type] = {}\n \n for sentiment in ['pos', 'neg']:\n data[data_type][sentiment] = []\n labels[data_type][sentiment] = []\n \n path = os.path.join(data_dir, data_type, sentiment, '*.txt')\n files = glob.glob(path)\n \n for f in files:\n with open(f) as review:\n data[data_type][sentiment].append(review.read())\n # Here we represent a positive review by '1' and a negative review by '0'\n labels[data_type][sentiment].append(1 if sentiment == 'pos' else 0)\n \n assert len(data[data_type][sentiment]) == len(labels[data_type][sentiment]), \\\n \"{}/{} data size does not match labels size\".format(data_type, sentiment)\n \n return data, labels", "_____no_output_____" ], [ "data, labels = read_imdb_data()\nprint(\"IMDB reviews: train = {} pos / {} neg, test = {} pos / {} neg\".format(\n len(data['train']['pos']), len(data['train']['neg']),\n len(data['test']['pos']), len(data['test']['neg'])))", "_____no_output_____" ] ], [ [ "Now that we've read the raw training and testing data from the downloaded dataset, we will combine the positive and negative reviews and shuffle the resulting records.", "_____no_output_____" ] ], [ [ "from sklearn.utils import shuffle\n\ndef prepare_imdb_data(data, labels):\n \"\"\"Prepare training and test sets from IMDb movie reviews.\"\"\"\n \n #Combine positive and negative reviews and labels\n data_train = data['train']['pos'] + data['train']['neg']\n data_test = data['test']['pos'] + data['test']['neg']\n labels_train = labels['train']['pos'] + labels['train']['neg']\n labels_test = labels['test']['pos'] + labels['test']['neg']\n \n #Shuffle reviews and corresponding labels within training and test sets\n data_train, labels_train = shuffle(data_train, labels_train)\n data_test, labels_test = shuffle(data_test, labels_test)\n \n # Return a unified training data, test data, training labels, test labets\n return data_train, data_test, labels_train, labels_test", "_____no_output_____" ], [ "train_X, test_X, train_y, test_y = prepare_imdb_data(data, labels)\nprint(\"IMDb reviews (combined): train = {}, test = {}\".format(len(train_X), len(test_X)))", "_____no_output_____" ] ], [ [ "Now that we have our training and testing sets unified and prepared, we should do a quick check and see an example of the data our model will be trained on. This is generally a good idea as it allows you to see how each of the further processing steps affects the reviews and it also ensures that the data has been loaded correctly.", "_____no_output_____" ] ], [ [ "print(train_X[100])\nprint(train_y[100])", "_____no_output_____" ] ], [ [ "The first step in processing the reviews is to make sure that any html tags that appear should be removed. In addition we wish to tokenize our input, that way words such as *entertained* and *entertaining* are considered the same with regard to sentiment analysis.", "_____no_output_____" ] ], [ [ "import nltk\nfrom nltk.corpus import stopwords\nfrom nltk.stem.porter import *\n\nimport re\nfrom bs4 import BeautifulSoup\n\ndef review_to_words(review):\n nltk.download(\"stopwords\", quiet=True)\n stemmer = PorterStemmer()\n \n text = BeautifulSoup(review, \"html.parser\").get_text() # Remove HTML tags\n text = re.sub(r\"[^a-zA-Z0-9]\", \" \", text.lower()) # Convert to lower case\n words = text.split() # Split string into words\n words = [w for w in words if w not in stopwords.words(\"english\")] # Remove stopwords\n words = [PorterStemmer().stem(w) for w in words] # stem\n \n return words", "_____no_output_____" ] ], [ [ "The `review_to_words` method defined above uses `BeautifulSoup` to remove any html tags that appear and uses the `nltk` package to tokenize the reviews. As a check to ensure we know how everything is working, try applying `review_to_words` to one of the reviews in the training set.", "_____no_output_____" ] ], [ [ "# TODO: Apply review_to_words to a review (train_X[100] or any other review)\nreview_to_words(train_X[100])", "_____no_output_____" ] ], [ [ "**Question:** Above we mentioned that `review_to_words` method removes html formatting and allows us to tokenize the words found in a review, for example, converting *entertained* and *entertaining* into *entertain* so that they are treated as though they are the same word. What else, if anything, does this method do to the input?", "_____no_output_____" ], [ "**Answer:**", "_____no_output_____" ], [ "The method below applies the `review_to_words` method to each of the reviews in the training and testing datasets. In addition it caches the results. This is because performing this processing step can take a long time. This way if you are unable to complete the notebook in the current session, you can come back without needing to process the data a second time.", "_____no_output_____" ] ], [ [ "import pickle\n\ncache_dir = os.path.join(\"../cache\", \"sentiment_analysis\") # where to store cache files\nos.makedirs(cache_dir, exist_ok=True) # ensure cache directory exists\n\ndef preprocess_data(data_train, data_test, labels_train, labels_test,\n cache_dir=cache_dir, cache_file=\"preprocessed_data.pkl\"):\n \"\"\"Convert each review to words; read from cache if available.\"\"\"\n\n # If cache_file is not None, try to read from it first\n cache_data = None\n if cache_file is not None:\n try:\n with open(os.path.join(cache_dir, cache_file), \"rb\") as f:\n cache_data = pickle.load(f)\n print(\"Read preprocessed data from cache file:\", cache_file)\n except:\n pass # unable to read from cache, but that's okay\n \n # If cache is missing, then do the heavy lifting\n if cache_data is None:\n # Preprocess training and test data to obtain words for each review\n #words_train = list(map(review_to_words, data_train))\n #words_test = list(map(review_to_words, data_test))\n words_train = [review_to_words(review) for review in data_train]\n words_test = [review_to_words(review) for review in data_test]\n \n # Write to cache file for future runs\n if cache_file is not None:\n cache_data = dict(words_train=words_train, words_test=words_test,\n labels_train=labels_train, labels_test=labels_test)\n with open(os.path.join(cache_dir, cache_file), \"wb\") as f:\n pickle.dump(cache_data, f)\n print(\"Wrote preprocessed data to cache file:\", cache_file)\n else:\n # Unpack data loaded from cache file\n words_train, words_test, labels_train, labels_test = (cache_data['words_train'],\n cache_data['words_test'], cache_data['labels_train'], cache_data['labels_test'])\n \n return words_train, words_test, labels_train, labels_test", "_____no_output_____" ], [ "# Preprocess data\ntrain_X, test_X, train_y, test_y = preprocess_data(train_X, test_X, train_y, test_y)", "_____no_output_____" ] ], [ [ "## Transform the data\n\nIn the XGBoost notebook we transformed the data from its word representation to a bag-of-words feature representation. For the model we are going to construct in this notebook we will construct a feature representation which is very similar. To start, we will represent each word as an integer. Of course, some of the words that appear in the reviews occur very infrequently and so likely don't contain much information for the purposes of sentiment analysis. The way we will deal with this problem is that we will fix the size of our working vocabulary and we will only include the words that appear most frequently. We will then combine all of the infrequent words into a single category and, in our case, we will label it as `1`.\n\nSince we will be using a recurrent neural network, it will be convenient if the length of each review is the same. To do this, we will fix a size for our reviews and then pad short reviews with the category 'no word' (which we will label `0`) and truncate long reviews.", "_____no_output_____" ], [ "### (TODO) Create a word dictionary\n\nTo begin with, we need to construct a way to map words that appear in the reviews to integers. Here we fix the size of our vocabulary (including the 'no word' and 'infrequent' categories) to be `5000` but you may wish to change this to see how it affects the model.\n\n> **TODO:** Complete the implementation for the `build_dict()` method below. Note that even though the vocab_size is set to `5000`, we only want to construct a mapping for the most frequently appearing `4998` words. This is because we want to reserve the special labels `0` for 'no word' and `1` for 'infrequent word'.", "_____no_output_____" ] ], [ [ "import numpy as np\n\ndef build_dict(data, vocab_size = 5000):\n \"\"\"Construct and return a dictionary mapping each of the most frequently appearing words to a unique integer.\"\"\"\n \n # TODO: Determine how often each word appears in `data`. Note that `data` is a list of sentences and that a\n # sentence is a list of words.\n \n word_count = {} # A dict storing the words that appear in the reviews along with how often they occur\n \n # TODO: Sort the words found in `data` so that sorted_words[0] is the most frequently appearing word and\n # sorted_words[-1] is the least frequently appearing word.\n \n sorted_words = None\n \n word_dict = {} # This is what we are building, a dictionary that translates words into integers\n for idx, word in enumerate(sorted_words[:vocab_size - 2]): # The -2 is so that we save room for the 'no word'\n word_dict[word] = idx + 2 # 'infrequent' labels\n \n return word_dict", "_____no_output_____" ], [ "word_dict = build_dict(train_X)", "_____no_output_____" ] ], [ [ "**Question:** What are the five most frequently appearing (tokenized) words in the training set? Does it makes sense that these words appear frequently in the training set?", "_____no_output_____" ], [ "**Answer:**", "_____no_output_____" ] ], [ [ "# TODO: Use this space to determine the five most frequently appearing words in the training set.", "_____no_output_____" ] ], [ [ "### Save `word_dict`\n\nLater on when we construct an endpoint which processes a submitted review we will need to make use of the `word_dict` which we have created. As such, we will save it to a file now for future use.", "_____no_output_____" ] ], [ [ "data_dir = '../data/pytorch' # The folder we will use for storing data\nif not os.path.exists(data_dir): # Make sure that the folder exists\n os.makedirs(data_dir)", "_____no_output_____" ], [ "with open(os.path.join(data_dir, 'word_dict.pkl'), \"wb\") as f:\n pickle.dump(word_dict, f)", "_____no_output_____" ] ], [ [ "### Transform the reviews\n\nNow that we have our word dictionary which allows us to transform the words appearing in the reviews into integers, it is time to make use of it and convert our reviews to their integer sequence representation, making sure to pad or truncate to a fixed length, which in our case is `500`.", "_____no_output_____" ] ], [ [ "def convert_and_pad(word_dict, sentence, pad=500):\n NOWORD = 0 # We will use 0 to represent the 'no word' category\n INFREQ = 1 # and we use 1 to represent the infrequent words, i.e., words not appearing in word_dict\n \n working_sentence = [NOWORD] * pad\n \n for word_index, word in enumerate(sentence[:pad]):\n if word in word_dict:\n working_sentence[word_index] = word_dict[word]\n else:\n working_sentence[word_index] = INFREQ\n \n return working_sentence, min(len(sentence), pad)\n\ndef convert_and_pad_data(word_dict, data, pad=500):\n result = []\n lengths = []\n \n for sentence in data:\n converted, leng = convert_and_pad(word_dict, sentence, pad)\n result.append(converted)\n lengths.append(leng)\n \n return np.array(result), np.array(lengths)", "_____no_output_____" ], [ "train_X, train_X_len = convert_and_pad_data(word_dict, train_X)\ntest_X, test_X_len = convert_and_pad_data(word_dict, test_X)", "_____no_output_____" ] ], [ [ "As a quick check to make sure that things are working as intended, check to see what one of the reviews in the training set looks like after having been processeed. Does this look reasonable? What is the length of a review in the training set?", "_____no_output_____" ] ], [ [ "# Use this cell to examine one of the processed reviews to make sure everything is working as intended.", "_____no_output_____" ] ], [ [ "**Question:** In the cells above we use the `preprocess_data` and `convert_and_pad_data` methods to process both the training and testing set. Why or why not might this be a problem?", "_____no_output_____" ], [ "**Answer:**", "_____no_output_____" ], [ "## Step 3: Upload the data to S3\n\nAs in the XGBoost notebook, we will need to upload the training dataset to S3 in order for our training code to access it. For now we will save it locally and we will upload to S3 later on.\n\n### Save the processed training dataset locally\n\nIt is important to note the format of the data that we are saving as we will need to know it when we write the training code. In our case, each row of the dataset has the form `label`, `length`, `review[500]` where `review[500]` is a sequence of `500` integers representing the words in the review.", "_____no_output_____" ] ], [ [ "import pandas as pd\n \npd.concat([pd.DataFrame(train_y), pd.DataFrame(train_X_len), pd.DataFrame(train_X)], axis=1) \\\n .to_csv(os.path.join(data_dir, 'train.csv'), header=False, index=False)", "_____no_output_____" ] ], [ [ "### Uploading the training data\n\n\nNext, we need to upload the training data to the SageMaker default S3 bucket so that we can provide access to it while training our model.", "_____no_output_____" ] ], [ [ "import sagemaker\n\nsagemaker_session = sagemaker.Session()\n\nbucket = sagemaker_session.default_bucket()\nprefix = 'sagemaker/sentiment_rnn'\n\nrole = sagemaker.get_execution_role()", "_____no_output_____" ], [ "input_data = sagemaker_session.upload_data(path=data_dir, bucket=bucket, key_prefix=prefix)", "_____no_output_____" ] ], [ [ "**NOTE:** The cell above uploads the entire contents of our data directory. This includes the `word_dict.pkl` file. This is fortunate as we will need this later on when we create an endpoint that accepts an arbitrary review. For now, we will just take note of the fact that it resides in the data directory (and so also in the S3 training bucket) and that we will need to make sure it gets saved in the model directory.", "_____no_output_____" ], [ "## Step 4: Build and Train the PyTorch Model\n\nIn the XGBoost notebook we discussed what a model is in the SageMaker framework. In particular, a model comprises three objects\n\n - Model Artifacts,\n - Training Code, and\n - Inference Code,\n \neach of which interact with one another. In the XGBoost example we used training and inference code that was provided by Amazon. Here we will still be using containers provided by Amazon with the added benefit of being able to include our own custom code.\n\nWe will start by implementing our own neural network in PyTorch along with a training script. For the purposes of this project we have provided the necessary model object in the `model.py` file, inside of the `train` folder. You can see the provided implementation by running the cell below.", "_____no_output_____" ] ], [ [ "!pygmentize train/model.py", "_____no_output_____" ] ], [ [ "The important takeaway from the implementation provided is that there are three parameters that we may wish to tweak to improve the performance of our model. These are the embedding dimension, the hidden dimension and the size of the vocabulary. We will likely want to make these parameters configurable in the training script so that if we wish to modify them we do not need to modify the script itself. We will see how to do this later on. To start we will write some of the training code in the notebook so that we can more easily diagnose any issues that arise.\n\nFirst we will load a small portion of the training data set to use as a sample. It would be very time consuming to try and train the model completely in the notebook as we do not have access to a gpu and the compute instance that we are using is not particularly powerful. However, we can work on a small bit of the data to get a feel for how our training script is behaving.", "_____no_output_____" ] ], [ [ "import torch\nimport torch.utils.data\n\n# Read in only the first 250 rows\ntrain_sample = pd.read_csv(os.path.join(data_dir, 'train.csv'), header=None, names=None, nrows=250)\n\n# Turn the input pandas dataframe into tensors\ntrain_sample_y = torch.from_numpy(train_sample[[0]].values).float().squeeze()\ntrain_sample_X = torch.from_numpy(train_sample.drop([0], axis=1).values).long()\n\n# Build the dataset\ntrain_sample_ds = torch.utils.data.TensorDataset(train_sample_X, train_sample_y)\n# Build the dataloader\ntrain_sample_dl = torch.utils.data.DataLoader(train_sample_ds, batch_size=50)", "_____no_output_____" ] ], [ [ "### (TODO) Writing the training method\n\nNext we need to write the training code itself. This should be very similar to training methods that you have written before to train PyTorch models. We will leave any difficult aspects such as model saving / loading and parameter loading until a little later.", "_____no_output_____" ] ], [ [ "def train(model, train_loader, epochs, optimizer, loss_fn, device):\n for epoch in range(1, epochs + 1):\n model.train()\n total_loss = 0\n for batch in train_loader: \n batch_X, batch_y = batch\n \n batch_X = batch_X.to(device)\n batch_y = batch_y.to(device)\n \n # TODO: Complete this train method to train the model provided.\n \n total_loss += loss.data.item()\n print(\"Epoch: {}, BCELoss: {}\".format(epoch, total_loss / len(train_loader)))", "_____no_output_____" ] ], [ [ "Supposing we have the training method above, we will test that it is working by writing a bit of code in the notebook that executes our training method on the small sample training set that we loaded earlier. The reason for doing this in the notebook is so that we have an opportunity to fix any errors that arise early when they are easier to diagnose.", "_____no_output_____" ] ], [ [ "import torch.optim as optim\nfrom train.model import LSTMClassifier\n\ndevice = torch.device(\"cuda\" if torch.cuda.is_available() else \"cpu\")\nmodel = LSTMClassifier(32, 100, 5000).to(device)\noptimizer = optim.Adam(model.parameters())\nloss_fn = torch.nn.BCELoss()\n\ntrain(model, train_sample_dl, 5, optimizer, loss_fn, device)", "_____no_output_____" ] ], [ [ "In order to construct a PyTorch model using SageMaker we must provide SageMaker with a training script. We may optionally include a directory which will be copied to the container and from which our training code will be run. When the training container is executed it will check the uploaded directory (if there is one) for a `requirements.txt` file and install any required Python libraries, after which the training script will be run.", "_____no_output_____" ], [ "### (TODO) Training the model\n\nWhen a PyTorch model is constructed in SageMaker, an entry point must be specified. This is the Python file which will be executed when the model is trained. Inside of the `train` directory is a file called `train.py` which has been provided and which contains most of the necessary code to train our model. The only thing that is missing is the implementation of the `train()` method which you wrote earlier in this notebook.\n\n**TODO**: Copy the `train()` method written above and paste it into the `train/train.py` file where required.\n\nThe way that SageMaker passes hyperparameters to the training script is by way of arguments. These arguments can then be parsed and used in the training script. To see how this is done take a look at the provided `train/train.py` file.", "_____no_output_____" ] ], [ [ "from sagemaker.pytorch import PyTorch\n\nestimator = PyTorch(entry_point=\"train.py\",\n source_dir=\"train\",\n role=role,\n framework_version='0.4.0',\n train_instance_count=1,\n train_instance_type='ml.p2.xlarge',\n hyperparameters={\n 'epochs': 10,\n 'hidden_dim': 200,\n })", "_____no_output_____" ], [ "estimator.fit({'training': input_data})", "_____no_output_____" ] ], [ [ "## Step 5: Testing the model\n\nAs mentioned at the top of this notebook, we will be testing this model by first deploying it and then sending the testing data to the deployed endpoint. We will do this so that we can make sure that the deployed model is working correctly.\n\n## Step 6: Deploy the model for testing\n\nNow that we have trained our model, we would like to test it to see how it performs. Currently our model takes input of the form `review_length, review[500]` where `review[500]` is a sequence of `500` integers which describe the words present in the review, encoded using `word_dict`. Fortunately for us, SageMaker provides built-in inference code for models with simple inputs such as this.\n\nThere is one thing that we need to provide, however, and that is a function which loads the saved model. This function must be called `model_fn()` and takes as its only parameter a path to the directory where the model artifacts are stored. This function must also be present in the python file which we specified as the entry point. In our case the model loading function has been provided and so no changes need to be made.\n\n**NOTE**: When the built-in inference code is run it must import the `model_fn()` method from the `train.py` file. This is why the training code is wrapped in a main guard ( ie, `if __name__ == '__main__':` )\n\nSince we don't need to change anything in the code that was uploaded during training, we can simply deploy the current model as-is.\n\n**NOTE:** When deploying a model you are asking SageMaker to launch an compute instance that will wait for data to be sent to it. As a result, this compute instance will continue to run until *you* shut it down. This is important to know since the cost of a deployed endpoint depends on how long it has been running for.\n\nIn other words **If you are no longer using a deployed endpoint, shut it down!**\n\n**TODO:** Deploy the trained model.", "_____no_output_____" ] ], [ [ "# TODO: Deploy the trained model", "_____no_output_____" ] ], [ [ "## Step 7 - Use the model for testing\n\nOnce deployed, we can read in the test data and send it off to our deployed model to get some results. Once we collect all of the results we can determine how accurate our model is.", "_____no_output_____" ] ], [ [ "test_X = pd.concat([pd.DataFrame(test_X_len), pd.DataFrame(test_X)], axis=1)", "_____no_output_____" ], [ "# We split the data into chunks and send each chunk seperately, accumulating the results.\n\ndef predict(data, rows=512):\n split_array = np.array_split(data, int(data.shape[0] / float(rows) + 1))\n predictions = np.array([])\n for array in split_array:\n predictions = np.append(predictions, predictor.predict(array))\n \n return predictions", "_____no_output_____" ], [ "predictions = predict(test_X.values)\npredictions = [round(num) for num in predictions]", "_____no_output_____" ], [ "from sklearn.metrics import accuracy_score\naccuracy_score(test_y, predictions)", "_____no_output_____" ] ], [ [ "**Question:** How does this model compare to the XGBoost model you created earlier? Why might these two models perform differently on this dataset? Which do *you* think is better for sentiment analysis?", "_____no_output_____" ], [ "**Answer:**", "_____no_output_____" ], [ "### (TODO) More testing\n\nWe now have a trained model which has been deployed and which we can send processed reviews to and which returns the predicted sentiment. However, ultimately we would like to be able to send our model an unprocessed review. That is, we would like to send the review itself as a string. For example, suppose we wish to send the following review to our model.", "_____no_output_____" ] ], [ [ "test_review = 'The simplest pleasures in life are the best, and this film is one of them. Combining a rather basic storyline of love and adventure this movie transcends the usual weekend fair with wit and unmitigated charm.'", "_____no_output_____" ] ], [ [ "The question we now need to answer is, how do we send this review to our model?\n\nRecall in the first section of this notebook we did a bunch of data processing to the IMDb dataset. In particular, we did two specific things to the provided reviews.\n - Removed any html tags and stemmed the input\n - Encoded the review as a sequence of integers using `word_dict`\n \nIn order process the review we will need to repeat these two steps.\n\n**TODO**: Using the `review_to_words` and `convert_and_pad` methods from section one, convert `test_review` into a numpy array `test_data` suitable to send to our model. Remember that our model expects input of the form `review_length, review[500]`.", "_____no_output_____" ] ], [ [ "# TODO: Convert test_review into a form usable by the model and save the results in test_data\ntest_data = None", "_____no_output_____" ] ], [ [ "Now that we have processed the review, we can send the resulting array to our model to predict the sentiment of the review.", "_____no_output_____" ] ], [ [ "predictor.predict(test_data)", "_____no_output_____" ] ], [ [ "Since the return value of our model is close to `1`, we can be certain that the review we submitted is positive.", "_____no_output_____" ], [ "### Delete the endpoint\n\nOf course, just like in the XGBoost notebook, once we've deployed an endpoint it continues to run until we tell it to shut down. Since we are done using our endpoint for now, we can delete it.", "_____no_output_____" ] ], [ [ "estimator.delete_endpoint()", "_____no_output_____" ] ], [ [ "## Step 6 (again) - Deploy the model for the web app\n\nNow that we know that our model is working, it's time to create some custom inference code so that we can send the model a review which has not been processed and have it determine the sentiment of the review.\n\nAs we saw above, by default the estimator which we created, when deployed, will use the entry script and directory which we provided when creating the model. However, since we now wish to accept a string as input and our model expects a processed review, we need to write some custom inference code.\n\nWe will store the code that we write in the `serve` directory. Provided in this directory is the `model.py` file that we used to construct our model, a `utils.py` file which contains the `review_to_words` and `convert_and_pad` pre-processing functions which we used during the initial data processing, and `predict.py`, the file which will contain our custom inference code. Note also that `requirements.txt` is present which will tell SageMaker what Python libraries are required by our custom inference code.\n\nWhen deploying a PyTorch model in SageMaker, you are expected to provide four functions which the SageMaker inference container will use.\n - `model_fn`: This function is the same function that we used in the training script and it tells SageMaker how to load our model.\n - `input_fn`: This function receives the raw serialized input that has been sent to the model's endpoint and its job is to de-serialize and make the input available for the inference code.\n - `output_fn`: This function takes the output of the inference code and its job is to serialize this output and return it to the caller of the model's endpoint.\n - `predict_fn`: The heart of the inference script, this is where the actual prediction is done and is the function which you will need to complete.\n\nFor the simple website that we are constructing during this project, the `input_fn` and `output_fn` methods are relatively straightforward. We only require being able to accept a string as input and we expect to return a single value as output. You might imagine though that in a more complex application the input or output may be image data or some other binary data which would require some effort to serialize.\n\n### (TODO) Writing inference code\n\nBefore writing our custom inference code, we will begin by taking a look at the code which has been provided.", "_____no_output_____" ] ], [ [ "!pygmentize serve/predict.py", "_____no_output_____" ] ], [ [ "As mentioned earlier, the `model_fn` method is the same as the one provided in the training code and the `input_fn` and `output_fn` methods are very simple and your task will be to complete the `predict_fn` method. Make sure that you save the completed file as `predict.py` in the `serve` directory.\n\n**TODO**: Complete the `predict_fn()` method in the `serve/predict.py` file.", "_____no_output_____" ], [ "### Deploying the model\n\nNow that the custom inference code has been written, we will create and deploy our model. To begin with, we need to construct a new PyTorchModel object which points to the model artifacts created during training and also points to the inference code that we wish to use. Then we can call the deploy method to launch the deployment container.\n\n**NOTE**: The default behaviour for a deployed PyTorch model is to assume that any input passed to the predictor is a `numpy` array. In our case we want to send a string so we need to construct a simple wrapper around the `RealTimePredictor` class to accomodate simple strings. In a more complicated situation you may want to provide a serialization object, for example if you wanted to sent image data.", "_____no_output_____" ] ], [ [ "from sagemaker.predictor import RealTimePredictor\nfrom sagemaker.pytorch import PyTorchModel\n\nclass StringPredictor(RealTimePredictor):\n def __init__(self, endpoint_name, sagemaker_session):\n super(StringPredictor, self).__init__(endpoint_name, sagemaker_session, content_type='text/plain')\n\nmodel = PyTorchModel(model_data=estimator.model_data,\n role = role,\n framework_version='0.4.0',\n entry_point='predict.py',\n source_dir='serve',\n predictor_cls=StringPredictor)\npredictor = model.deploy(initial_instance_count=1, instance_type='ml.m4.xlarge')", "_____no_output_____" ] ], [ [ "### Testing the model\n\nNow that we have deployed our model with the custom inference code, we should test to see if everything is working. Here we test our model by loading the first `250` positive and negative reviews and send them to the endpoint, then collect the results. The reason for only sending some of the data is that the amount of time it takes for our model to process the input and then perform inference is quite long and so testing the entire data set would be prohibitive.", "_____no_output_____" ] ], [ [ "import glob\n\ndef test_reviews(data_dir='../data/aclImdb', stop=250):\n \n results = []\n ground = []\n \n # We make sure to test both positive and negative reviews \n for sentiment in ['pos', 'neg']:\n \n path = os.path.join(data_dir, 'test', sentiment, '*.txt')\n files = glob.glob(path)\n \n files_read = 0\n \n print('Starting ', sentiment, ' files')\n \n # Iterate through the files and send them to the predictor\n for f in files:\n with open(f) as review:\n # First, we store the ground truth (was the review positive or negative)\n if sentiment == 'pos':\n ground.append(1)\n else:\n ground.append(0)\n # Read in the review and convert to 'utf-8' for transmission via HTTP\n review_input = review.read().encode('utf-8')\n # Send the review to the predictor and store the results\n results.append(float(predictor.predict(review_input)))\n \n # Sending reviews to our endpoint one at a time takes a while so we\n # only send a small number of reviews\n files_read += 1\n if files_read == stop:\n break\n \n return ground, results", "_____no_output_____" ], [ "ground, results = test_reviews()", "_____no_output_____" ], [ "from sklearn.metrics import accuracy_score\naccuracy_score(ground, results)", "_____no_output_____" ] ], [ [ "As an additional test, we can try sending the `test_review` that we looked at earlier.", "_____no_output_____" ] ], [ [ "predictor.predict(test_review)", "_____no_output_____" ] ], [ [ "Now that we know our endpoint is working as expected, we can set up the web page that will interact with it. If you don't have time to finish the project now, make sure to skip down to the end of this notebook and shut down your endpoint. You can deploy it again when you come back.", "_____no_output_____" ], [ "## Step 7 (again): Use the model for the web app\n\n> **TODO:** This entire section and the next contain tasks for you to complete, mostly using the AWS console.\n\nSo far we have been accessing our model endpoint by constructing a predictor object which uses the endpoint and then just using the predictor object to perform inference. What if we wanted to create a web app which accessed our model? The way things are set up currently makes that not possible since in order to access a SageMaker endpoint the app would first have to authenticate with AWS using an IAM role which included access to SageMaker endpoints. However, there is an easier way! We just need to use some additional AWS services.\n\n<img src=\"Web App Diagram.svg\">\n\nThe diagram above gives an overview of how the various services will work together. On the far right is the model which we trained above and which is deployed using SageMaker. On the far left is our web app that collects a user's movie review, sends it off and expects a positive or negative sentiment in return.\n\nIn the middle is where some of the magic happens. We will construct a Lambda function, which you can think of as a straightforward Python function that can be executed whenever a specified event occurs. We will give this function permission to send and recieve data from a SageMaker endpoint.\n\nLastly, the method we will use to execute the Lambda function is a new endpoint that we will create using API Gateway. This endpoint will be a url that listens for data to be sent to it. Once it gets some data it will pass that data on to the Lambda function and then return whatever the Lambda function returns. Essentially it will act as an interface that lets our web app communicate with the Lambda function.\n\n### Setting up a Lambda function\n\nThe first thing we are going to do is set up a Lambda function. This Lambda function will be executed whenever our public API has data sent to it. When it is executed it will receive the data, perform any sort of processing that is required, send the data (the review) to the SageMaker endpoint we've created and then return the result.\n\n#### Part A: Create an IAM Role for the Lambda function\n\nSince we want the Lambda function to call a SageMaker endpoint, we need to make sure that it has permission to do so. To do this, we will construct a role that we can later give the Lambda function.\n\nUsing the AWS Console, navigate to the **IAM** page and click on **Roles**. Then, click on **Create role**. Make sure that the **AWS service** is the type of trusted entity selected and choose **Lambda** as the service that will use this role, then click **Next: Permissions**.\n\nIn the search box type `sagemaker` and select the check box next to the **AmazonSageMakerFullAccess** policy. Then, click on **Next: Review**.\n\nLastly, give this role a name. Make sure you use a name that you will remember later on, for example `LambdaSageMakerRole`. Then, click on **Create role**.\n\n#### Part B: Create a Lambda function\n\nNow it is time to actually create the Lambda function.\n\nUsing the AWS Console, navigate to the AWS Lambda page and click on **Create a function**. When you get to the next page, make sure that **Author from scratch** is selected. Now, name your Lambda function, using a name that you will remember later on, for example `sentiment_analysis_func`. Make sure that the **Python 3.6** runtime is selected and then choose the role that you created in the previous part. Then, click on **Create Function**.\n\nOn the next page you will see some information about the Lambda function you've just created. If you scroll down you should see an editor in which you can write the code that will be executed when your Lambda function is triggered. In our example, we will use the code below. \n\n```python\n# We need to use the low-level library to interact with SageMaker since the SageMaker API\n# is not available natively through Lambda.\nimport boto3\n\ndef lambda_handler(event, context):\n\n # The SageMaker runtime is what allows us to invoke the endpoint that we've created.\n runtime = boto3.Session().client('sagemaker-runtime')\n\n # Now we use the SageMaker runtime to invoke our endpoint, sending the review we were given\n response = runtime.invoke_endpoint(EndpointName = '**ENDPOINT NAME HERE**', # The name of the endpoint we created\n ContentType = 'text/plain', # The data format that is expected\n Body = event['body']) # The actual review\n\n # The response is an HTTP response whose body contains the result of our inference\n result = response['Body'].read().decode('utf-8')\n\n return {\n 'statusCode' : 200,\n 'headers' : { 'Content-Type' : 'text/plain', 'Access-Control-Allow-Origin' : '*' },\n 'body' : result\n }\n```\n\nOnce you have copy and pasted the code above into the Lambda code editor, replace the `**ENDPOINT NAME HERE**` portion with the name of the endpoint that we deployed earlier. You can determine the name of the endpoint using the code cell below.", "_____no_output_____" ] ], [ [ "predictor.endpoint", "_____no_output_____" ] ], [ [ "Once you have added the endpoint name to the Lambda function, click on **Save**. Your Lambda function is now up and running. Next we need to create a way for our web app to execute the Lambda function.\n\n### Setting up API Gateway\n\nNow that our Lambda function is set up, it is time to create a new API using API Gateway that will trigger the Lambda function we have just created.\n\nUsing AWS Console, navigate to **Amazon API Gateway** and then click on **Get started**.\n\nOn the next page, make sure that **New API** is selected and give the new api a name, for example, `sentiment_analysis_api`. Then, click on **Create API**.\n\nNow we have created an API, however it doesn't currently do anything. What we want it to do is to trigger the Lambda function that we created earlier.\n\nSelect the **Actions** dropdown menu and click **Create Method**. A new blank method will be created, select its dropdown menu and select **POST**, then click on the check mark beside it.\n\nFor the integration point, make sure that **Lambda Function** is selected and click on the **Use Lambda Proxy integration**. This option makes sure that the data that is sent to the API is then sent directly to the Lambda function with no processing. It also means that the return value must be a proper response object as it will also not be processed by API Gateway.\n\nType the name of the Lambda function you created earlier into the **Lambda Function** text entry box and then click on **Save**. Click on **OK** in the pop-up box that then appears, giving permission to API Gateway to invoke the Lambda function you created.\n\nThe last step in creating the API Gateway is to select the **Actions** dropdown and click on **Deploy API**. You will need to create a new Deployment stage and name it anything you like, for example `prod`.\n\nYou have now successfully set up a public API to access your SageMaker model. Make sure to copy or write down the URL provided to invoke your newly created public API as this will be needed in the next step. This URL can be found at the top of the page, highlighted in blue next to the text **Invoke URL**.", "_____no_output_____" ], [ "## Step 4: Deploying our web app\n\nNow that we have a publicly available API, we can start using it in a web app. For our purposes, we have provided a simple static html file which can make use of the public api you created earlier.\n\nIn the `website` folder there should be a file called `index.html`. Download the file to your computer and open that file up in a text editor of your choice. There should be a line which contains **\\*\\*REPLACE WITH PUBLIC API URL\\*\\***. Replace this string with the url that you wrote down in the last step and then save the file.\n\nNow, if you open `index.html` on your local computer, your browser will behave as a local web server and you can use the provided site to interact with your SageMaker model.\n\nIf you'd like to go further, you can host this html file anywhere you'd like, for example using github or hosting a static site on Amazon's S3. Once you have done this you can share the link with anyone you'd like and have them play with it too!\n\n> **Important Note** In order for the web app to communicate with the SageMaker endpoint, the endpoint has to actually be deployed and running. This means that you are paying for it. Make sure that the endpoint is running when you want to use the web app but that you shut it down when you don't need it, otherwise you will end up with a surprisingly large AWS bill.\n\n**TODO:** Make sure that you include the edited `index.html` file in your project submission.", "_____no_output_____" ], [ "Now that your web app is working, trying playing around with it and see how well it works.\n\n**Question**: Give an example of a review that you entered into your web app. What was the predicted sentiment of your example review?", "_____no_output_____" ], [ "**Answer:**", "_____no_output_____" ], [ "### Delete the endpoint\n\nRemember to always shut down your endpoint if you are no longer using it. You are charged for the length of time that the endpoint is running so if you forget and leave it on you could end up with an unexpectedly large bill.", "_____no_output_____" ] ], [ [ "predictor.delete_endpoint()", "_____no_output_____" ] ] ]

[ "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code" ]

[ [ "markdown" ], [ "code" ], [ "markdown" ], [ "code", "code" ], [ "markdown" ], [ "code", "code" ], [ "markdown" ], [ "code", "code" ], [ "markdown" ], [ "code" ], [ "markdown" ], [ "code" ], [ "markdown" ], [ "code" ], [ "markdown", "markdown", "markdown" ], [ "code", "code" ], [ "markdown", "markdown" ], [ "code", "code" ], [ "markdown", "markdown" ], [ "code" ], [ "markdown" ], [ "code", "code" ], [ "markdown" ], [ "code", "code" ], [ "markdown" ], [ "code" ], [ "markdown", "markdown", "markdown" ], [ "code" ], [ "markdown" ], [ "code", "code" ], [ "markdown", "markdown" ], [ "code" ], [ "markdown" ], [ "code" ], [ "markdown" ], [ "code" ], [ "markdown" ], [ "code" ], [ "markdown", "markdown" ], [ "code", "code" ], [ "markdown" ], [ "code" ], [ "markdown" ], [ "code", "code", "code", "code" ], [ "markdown", "markdown", "markdown" ], [ "code" ], [ "markdown" ], [ "code" ], [ "markdown" ], [ "code" ], [ "markdown", "markdown" ], [ "code" ], [ "markdown" ], [ "code" ], [ "markdown", "markdown" ], [ "code" ], [ "markdown" ], [ "code", "code", "code" ], [ "markdown" ], [ "code" ], [ "markdown", "markdown" ], [ "code" ], [ "markdown", "markdown", "markdown", "markdown", "markdown" ], [ "code" ] ]

[ "MIT" ]

[ "MIT" ]

[ "MIT" ]

[ [ [ "# Lesson 03 - Central Tendency\n\nWe have distributions and we need some numebr to describe that data.\n- Mode\n - For numbers it is single number\n - For Distributions it is range of numbers\n - Some distributions don't have mode e.g. uniform distribution\n - Some distributions have multiple modes \n - it is not affected by outliers\n- Median\n - median is the [actual average of people's intuition](http://www.slate.com/articles/life/weddings/2013/06/average_wedding_cost_published_numbers_on_the_price_of_a_wedding_are_totally.html)\n - median is more robust to outliers\n- Mean\n - $\\bar{x} = (\\sum x) / n$\n - $\\mu = (\\sum x) / N$\n - All scores affect the mean\n - many samples from same population will have same mean\n - it is affected by outliers\n \n", "_____no_output_____" ] ] ]

[ "markdown" ]

[ [ "markdown" ] ]

[ "Apache-2.0" ]

[ "Apache-2.0" ]

[ "Apache-2.0" ]

[ [ [ "# Probability and statistics\n\n\nIn some form or another, machine learning is all about making predictions. \nWe might want to predict the *probability* of a patient suffering a heart attack in the next year,\ngiven their clinical history.\nIn anomaly detection, we might want to assess how *likely* a set of readings from an airplane's jet engine would be,\nwere it operating normally. \nIn reinforcement learning, we want an agent to act intelligently in an environment. \nThis means we need to think about the probability of getting a high reward under each of the available action. \nAnd when we build recommender systems we also need to think about probability. \nFor example, if we *hypothetically* worked for a large online bookseller,\nwe might want to estimate the probability that a particular user would buy a particular book, if prompted. \nFor this we need to use the language of probability and statistics. \nEntire courses, majors, theses, careers, and even departments, are devoted to probability.\nSo our goal here isn't to teach the whole subject. \nInstead we hope to get you off the ground,\nto teach you just enough that you know everything necessary to start building your first machine learning models \nand to have enough of a flavor for the subject that you can begin to explore it on your own if you wish.\n\n\nWe've talked a lot about probabilities so far without articulating what precisely they are or giving a concrete example. Let's get more serious by considering the problem of distinguishing cats and dogs based on photographs. This might sound simpler but it's actually a formidable challenge. To start with, the difficulty of the problem may depend on the resolution of the image.\n\n| 20px | 40px | 80px | 160px | 320px |\n|:----:|:----:|:----:|:-----:|:-----:|\n||||||\n||||||\n\nWhile it's easy for humans to recognize cats and dogs at 320 pixel resolution, \nit becomes challenging at 40 pixels \nand next to impossible at 20 pixels. \nIn other words, our ability to tell cats and dogs apart at a large distance (and thus low resolution) \nmight approach uninformed guessing. \nProbability gives us a formal way of reasoning about our level of certainty.\nIf we are completely sure that the image depicts a cat, \nwe say that the *probability* that the corresponding label $l$ is $\\mathrm{cat}$, \ndenoted $P(l=\\mathrm{cat})$ equals 1.0.\nIf we had no evidence to suggest that $l =\\mathrm{cat}$ or that $l = \\mathrm{dog}$,\nthen we might say that the two possibilities were equally $likely$\nexpressing this as $P(l=\\mathrm{cat}) = 0.5$.\nIf we were reasonably confident, but not sure that the image depicted a cat,\nwe might assign a probability $.5 < P(l=\\mathrm{cat}) < 1.0$.\n\nNow consider a second case:\ngiven some weather monitoring data,\nwe want to predict the probability that it will rain in Taipei tomorrow.\nIf it's summertime, the rain might come with probability $.5$\nIn both cases, we have some value of interest.\nAnd in both cases we are uncertain about the outcome.\nBut There's a key difference between the two cases. \nIn this first case, the image is in fact either a dog or a cat, \nwe just don't know which. \nIn the second case, the outcome may actually be a random event,\nif you believe in such things (and most physicists do).\nSo probability is a flexible language for reasoning about our level of certainty,\nand it can be applied effectively in a broad set of contexts.", "_____no_output_____" ], [ "## Basic probability theory\n\nSay that we cast a die and want to know \nwhat the chance is of seeing a $1$ \nrather than another digit. \nIf the die is fair, all six outcomes $\\mathcal{X} = \\{1, \\ldots, 6\\}$ \nare equally likely to occur, \nhence we would see a $1$ in $1$ out of $6$ cases. \nFormally we state that $1$ occurs with probability $\\frac{1}{6}$. \n\nFor a real die that we receive from a factory,\nwe might not know those proportions \nand we would need to check whether it is tainted. \nThe only way to investigate the die is by casting it many times \nand recording the outcomes. \nFor each cast of the die, \nwe'll observe a value $\\{1, 2, \\ldots, 6\\}$. \nGiven these outcomes, we want to investigate the probability of observing each outcome.\n\nOne natural approach for each value is to take the individual count for that value\nand to divide it by the total number of tosses.\nThis gives us an *estimate* of the probability of a given event. \nThe law of large numbers tell us that as the number of tosses grows this estimate will draw closer and closer to the true underlying probability.\nBefore going into the details of what's going here, let's try it out.\n\nTo start, let's import the necessary packages:", "_____no_output_____" ] ], [ [ "import mxnet as mx\nfrom mxnet import nd", "_____no_output_____" ] ], [ [ "Next, we'll want to be able to cast the die.\nIn statistics we call this process of drawing examples from probability distributions *sampling*.\nThe distribution which assigns probabilities to a number of discrete choices is called \nthe *multinomial* distribution. \nWe'll give a more formal definition of *distribution* later,\nbut at a high level, think of it as just an assignment of probabilities to events.\nIn MXNet, we can sample from the multinomial distribution via the aptly named `nd.sample_multinomial` function.\nThe function can be called in many ways, but we'll focus on the simplest.\nTo draw a single sample, we simply give pass in a vector of probabilities.", "_____no_output_____" ] ], [ [ "probabilities = nd.ones(6) / 6\nnd.sample_multinomial(probabilities)", "_____no_output_____" ] ], [ [ "If you run this line (`nd.sample_multinomial(probabilities)`) a bunch of times, \nyou'll find that you get out random values each time. \nAs with estimating the fairness of a die, \nwe often want to generate many samples from the same distribution.\nIt would be really slow to do this with a Python `for` loop,\nso `sample_multinomial` supports drawing multiple samples at once,\nreturning an array of independent samples in any shape we might desire.", "_____no_output_____" ] ], [ [ "print(nd.sample_multinomial(probabilities, shape=(10)))\nprint(nd.sample_multinomial(probabilities, shape=(5,10)))", "\n[3 4 5 3 5 3 5 2 3 3]\n<NDArray 10 @cpu(0)>\n\n[[2 2 1 5 0 5 1 2 2 4]\n [4 3 2 3 2 5 5 0 2 0]\n [3 0 2 4 5 4 0 5 5 5]\n [2 4 4 2 3 4 4 0 4 3]\n [3 0 3 5 4 3 0 2 2 1]]\n<NDArray 5x10 @cpu(0)>\n" ] ], [ [ "Now that we know how to sample rolls of a die,\nwe can simulate 1000 rolls.", "_____no_output_____" ] ], [ [ "rolls = nd.sample_multinomial(probabilities, shape=(1000))", "_____no_output_____" ] ], [ [ "We can then go through and count, after each of the 1000 rolls,\nhow many times each number was rolled.", "_____no_output_____" ] ], [ [ "counts = nd.zeros((6,1000))\ntotals = nd.zeros(6)\nfor i, roll in enumerate(rolls):\n totals[int(roll.asscalar())] += 1\n counts[:, i] = totals", "_____no_output_____" ] ], [ [ "To start, we can inspect the final tally at the end of $1000$ rolls.", "_____no_output_____" ] ], [ [ "totals / 1000", "_____no_output_____" ] ], [ [ "As you can see, the lowest estimated probability for any of the numbers is about $.15$ \nand the highest estimated probability is $0.188$. \nBecause we generated the data from a fair die,\nwe know that each number actually has probability of $1/6$, roughly $.167$,\nso these estimates are pretty good. \nWe can also visualize how these probabilities converge over time\ntowards reasonable estimates.\n\nTo start let's take a look at the `counts`\narray which has shape `(6, 1000)`.\nFor each time step (out of 1000),\ncounts, says how many times each of the numbers has shown up.\nSo we can normalize each $j$-th column of the counts vector by the number of tosses\nto give the `current` estimated probabilities at that time.\nThe counts object looks like this:", "_____no_output_____" ] ], [ [ "counts", "_____no_output_____" ] ], [ [ "Normalizing by the number of tosses, we get:", "_____no_output_____" ] ], [ [ "x = nd.arange(1000).reshape((1,1000)) + 1\nestimates = counts / x\nprint(estimates[:,0])\nprint(estimates[:,1])\nprint(estimates[:,100])", "\n[ 0. 1. 0. 0. 0. 0.]\n<NDArray 6 @cpu(0)>\n\n[ 0. 0.5 0. 0. 0.5 0. ]\n<NDArray 6 @cpu(0)>\n\n[ 0.1980198 0.15841584 0.17821783 0.18811882 0.12871288 0.14851485]\n<NDArray 6 @cpu(0)>\n" ] ], [ [ "As you can see, after the first toss of the die, we get the extreme estimate that one of the numbers will be rolled with probability $1.0$ and that the others have probability $0$. After $100$ rolls, things already look a bit more reasonable.\nWe can visualize this convergence by using the plotting package `matplotlib`. \nIf you don't have it installed, now would be a good time to [install it](https://matplotlib.org/).\n", "_____no_output_____" ] ], [ [ "from matplotlib import pyplot as plt\nplt.plot(estimates[0, :].asnumpy(), label=\"Estimated P(die=1)\")\nplt.plot(estimates[1, :].asnumpy(), label=\"Estimated P(die=2)\")\nplt.plot(estimates[2, :].asnumpy(), label=\"Estimated P(die=3)\")\nplt.plot(estimates[3, :].asnumpy(), label=\"Estimated P(die=4)\")\nplt.plot(estimates[4, :].asnumpy(), label=\"Estimated P(die=5)\")\nplt.plot(estimates[5, :].asnumpy(), label=\"Estimated P(die=6)\")\nplt.axhline(y=0.16666, color='black', linestyle='dashed')\nplt.legend()\nplt.show()", "_____no_output_____" ] ], [ [ "Each solid curve corresponds to one of the six values of the die\nand gives our estimated probability that the die turns up that value \nas assessed after each of the 1000 turns. \nThe dashed black line gives the true underlying probability.\nAs we get more data, the solid curves converge towards the true answer.", "_____no_output_____" ], [ "<!-- What we can see is that the red curves pretty well capture the behavior of the 10 random traces of averages. This is the case since we are averaging numbers and their aggregate behavior is like that of a number with a lot less uncertainty. Looking at the red curves, they are given by $f(x) = \\pm 1/\\sqrt{x}$. (The reader might cry foul by noting that we just added Gaussian random variables which, quite obviously, lead to yet another Gaussian random variable. That said, the curves for sums of other random variables, such as $\\{0, 1\\}$ valued objects look identical in the limit.) -->\n\n\nIn our example of casting a die, we introduced the notion of a **random variable**. \nA random variable, which we denote here as $X$ can be pretty much any quantity is not determistic. \nRandom variables could take one value among a set of possibilites. \nWe denote sets with brackets, e.g., $\\{\\mathrm{cat}, \\mathrm{dog}, \\mathrm{rabbit}\\}$.\nThe items contained in the set are called *elements*,\nand we can say that an element $x$ is *in* the set S, by writing $x \\in S$.\nThe symbol $\\in$ is read as \"in\" and denotes membership. \nFor instance, we could truthfully say $\\mathrm{dog} \\in \\{\\mathrm{cat}, \\mathrm{dog}, \\mathrm{rabbit}\\}$.\nWhen dealing with the rolls of die, we are concerned with a variable $X \\in \\{1, 2, 3, 4, 5, 6\\}$. \n\nNote that there is a subtle difference between discrete random variables, like the sides of a dice, \nand continuous ones, like the weight and the height of a person. \nThere's little point in asking whether two people have exactly the same height. \nIf we take precise enough measurements you'll find that no two people on the planet have the exact same height. \nIn fact, if we take a fine enough measurement, \nyou will not have the same height when you wake up and when you go to sleep.\nSo there's no purpose in asking about the probability \nthat some one is $2.00139278291028719210196740527486202$ meters tall. \nThe probability is 0.\nIt makes more sense in this case to ask whether someone's height falls into a given interval, \nsay between 1.99 and 2.01 meters. \nIn these cases we quantify the likelihood that we see a value as a density. \nThe height of exactly 2.0 meters has no probability, but nonzero density. \nBetween any two different heights we have nonzero probability.\n\n\nThere are a few important axioms of probability that you'll want to remember:\n\n* For any event $z$, the probability is never negative, i.e. $\\Pr(Z=z) \\geq 0$.\n* For any two events $Z=z$ and $X=x$ the union is no more likely than the sum of the individual events, i.e. $\\Pr(Z=z \\cup X=x) \\leq \\Pr(Z=z) + \\Pr(X=x)$.\n* For any random variable, the probabilities of all the values it can take must sum to 1 $\\sum_{i=1}^n P(Z=z_i) = 1$.\n* For any two mutually exclusive events $Z=z$ and $X=x$, the probability that either happens is equal to the sum of their individual probabilities that $\\Pr(Z=z \\cup X=x) = \\Pr(Z=z) + \\Pr(X=z)$.", "_____no_output_____" ], [ "## Dealing with multiple random variables\n\nVery often, we'll want consider more than one random variable at a time. \nFor instance, we may want to model the relationship between diseases and symptoms.\nGiven a disease and symptom, say 'flu' and 'cough', \neither may or may not occur in a patient with some probability.\nWhile we hope that the probability of both would be close to zero,\nwe may want to estimate these probabilities and their relationships to each other\nso that we may apply our inferences to effect better medical care.\n\nAs a more complicated example, images contain millions of pixels, thus millions of random variables. \nAnd in many cases images will come with a label, identifying objects in the image.\nWe can also think of the label as a random variable.\nWe can even get crazy and think of all the metadata as random variables\nsuch as location, time, aperture, focal length, ISO, focus distance, camera type, etc.\nAll of these are random variables that occur jointly. \nWhen we deal with multiple random variables, \nthere are several quantities of interest.\nThe first is called the joint distribution $\\Pr(A, B)$. \nGiven any elements $a$ and $b$,\nthe joint distribution lets us answer,\nwhat is the probability that $A=a$ and $B=b$ simulataneously?\nIt might be clear that for any values $a$ and $b$, $\\Pr(A,B) \\leq \\Pr(A=a)$. \n\nThis has to be the case, since for $A$ and $B$ to happen, \n$A$ has to happen *and* $B$ also has to happen (and vice versa). \nThus $A,B$ cannot be more likely than $A$ or $B$ individually. \nThis brings us to an interesting ratio: $0 \\leq \\frac{\\Pr(A,B)}{\\Pr(A)} \\leq 1$. \nWe call this a **conditional probability** and denote it by $\\Pr(B|A)$, \nthe probability that $B$ happens, provided that $A$ has happened. \n\nUsing the definition of conditional probabilities, \nwe can derive one of the most useful and celebrated equations in statistics - Bayes' theorem. \nIt goes as follows: By construction, we have that $\\Pr(A, B) = \\Pr(B|A) \\Pr(A)$. \nBy symmetry, this also holds for $\\Pr(A,B) = \\Pr(A|B) \\Pr(B)$. \nSolving for one of the conditional variables we get:\n$$\\Pr(A|B) = \\frac{\\Pr(B|A) \\Pr(A)}{\\Pr(B)}$$\n\nThis is very useful if we want to infer one thing from another, \nsay cause and effect but we only know the properties in the reverse direction. \nOne important operation that we need to make this work is **marginalization**, i.e., \nthe operation of determining $\\Pr(A)$ and $\\Pr(B)$ from $\\Pr(A,B)$.\nWe can see that the probability of seeing $A$ amounts to accounting \nfor all possible choices of $B$ and aggregating the joint probabilities over all of them, i.e. \n\n$$\\Pr(A) = \\sum_{B'} \\Pr(A,B') \\text{ and } \\Pr(B) = \\sum_{A'} \\Pr(A',B)$$\n\nA really useful property to check is for **dependence** and **independence**. \nIndependence is when the occurrence of one event does not influence the occurrence of the other.\nIn this case $\\Pr(B|A) = \\Pr(B)$. Statisticians typically use $A \\perp\\!\\!\\!\\perp B$ to express this. \nFrom Bayes Theorem it follows immediately that also $\\Pr(A|B) = \\Pr(A)$. \nIn all other cases we call $A$ and $B$ dependent. \nFor instance, two successive rolls of a dice are independent. \nOn the other hand, the position of a light switch and the brightness in the room are not \n(they are not perfectly deterministic, though, \nsince we could always have a broken lightbulb, power failure, or a broken switch). \n\nLet's put our skills to the test. \nAssume that a doctor administers an AIDS test to a patient. \nThis test is fairly accurate and fails only with 1% probability \nif the patient is healthy by reporting him as diseased, \nand that it never fails to detect HIV if the patient actually has it. \nWe use $D$ to indicate the diagnosis and $H$ to denote the HIV status.\nWritten as a table the outcome $\\Pr(D|H)$ looks as follows:\n\n| | Patient is HIV positive | Patient is HIV negative |\n|:------------|------------------------:|------------------------:|\n|Test positive| 1 | 0.01 |\n|Test negative| 0 | 0.99 |\n\nNote that the column sums are all one (but the row sums aren't), \nsince the conditional probability needs to sum up to $1$, just like the probability. \nLet us work out the probability of the patient having AIDS if the test comes back positive. \nObviously this is going to depend on how common the disease is, since it affects the number of false alarms.\nAssume that the population is quite healthy, e.g. $\\Pr(\\text{HIV positive}) = 0.0015$. \nTo apply Bayes Theorem we need to determine \n\n$$\\Pr(\\text{Test positive}) = \\Pr(D=1|H=0) \\Pr(H=0) + \\Pr(D=1|H=1) \\Pr(H=1) = 0.01 \\cdot 0.9985 + 1 \\cdot 0.0015 = 0.011485$$\n\nHence we get $\\Pr(H = 1|D = 1) = \\frac{\\Pr(D=1|H=1) \\Pr(H=1)}{\\Pr(D=1)} = \\frac{1 \\cdot 0.0015}{0.011485} = 0.131$, in other words, there's only a 13.1% chance that the patient actually has AIDS, despite using a test that is 99% accurate! As we can see, statistics can be quite counterintuitive. ", "_____no_output_____" ], [ "## Conditional independence\n\nWhat should a patient do upon receiving such terrifying news? \nLikely, he/she would ask the physician to administer another test to get clarity. \nThe second test has different characteristics (it isn't as good as the first one). \n\n| | Patient is HIV positive | Patient is HIV negative |\n|:------------|------------------------:|------------------------:|\n|Test positive| 0.98 | 0.03 |\n|Test negative| 0.02 | 0.97 |\n\nUnfortunately, the second test comes back positive, too. \nLet us work out the requisite probabilities to invoke Bayes' Theorem. \n\n* $\\Pr(D_1 = 1 \\text{ and } D_2 = 1|H = 0) = 0.01 \\cdot 0.03 = 0.0001$\n* $\\Pr(D_1 = 1 \\text{ and } D_2 = 1|H = 1) = 1 \\cdot 0.98 = 0.98$\n* $\\Pr(D_1 = 1 \\text{ and } D_2 = 1) = 0.0001 \\cdot 0.9985 + 0.98 \\cdot 0.0015 = 0.00156985$\n* $\\Pr(H = 1|D_1 = 1 \\text{ and } D_2 = 1) = \\frac{0.98 \\cdot 0.0015}{0.00156985} = 0.936$\n\nThat is, the second test allowed us to gain much higher confidence that not all is well. \nDespite the second test being considerably less accurate than the first one, \nit still improved our estimate quite a bit. \n*Why couldn't we just run the first test a second time?* \nAfter all, the first test was more accurate. \nThe reason is that we needed a second test that confirmed *independently* of the first test that things were dire, indeed. In other words, we made the tacit assumption that $\\Pr(D_1, D_2|H) = \\Pr(D_1|H) \\Pr(D_2|H)$. Statisticians call such random variables **conditionally independent**. This is expressed as $D_1 \\perp\\!\\!\\!\\perp D_2 | H$. \n\n## Naive Bayes classification\n\nConditional independence is useful when dealing with data, since it simplifies a lot of equations. \nA popular algorithm is the Naive Bayes Classifier. \nThe key assumption in it is that the attributes are all independent of each other, given the labels. \nIn other words, we have:\n\n$$p(x|y) = \\prod_i p(x_i|y)$$\n\nUsing Bayes Theorem this leads to the classifier $p(y|x) = \\frac{\\prod_i p(x_i|y) p(y)}{p(x)}$. Unfortunately, this is still intractable, since we don't know $p(x)$. Fortunately, we don't need it, since we know that $\\sum_y p(y|x) = 1$, hence we can always recover the normalization from $p(y|x) \\propto \\prod_i p(x_i|y) p(y)$. After all that math, it's time for some code to show how to use a Naive Bayes classifier for distinguishing digits on the MNIST classification dataset. \n\nThe problem is that we don't actually know $p(y)$ and $p(x_i|y)$. So we need to *estimate* it given some training data first. This is what is called *training* the model. In the case of 10 possible classes we simply compute $n_y$, i.e. the number of occurrences of class $y$ and then divide it by the total number of occurrences. E.g. if we have a total of 60,000 pictures of digits and digit 4 occurs 5800 times, we estimate its probability as $\\frac{5800}{60000}$. Likewise, to get an idea of $p(x_i|y)$ we count how many times pixel $i$ is set for digit $y$ and then divide it by the number of occurrences of digit $y$. This is the probability that that very pixel will be switched on.", "_____no_output_____" ] ], [ [ "import numpy as np\n\n# we go over one observation at a time (speed doesn't matter here)\ndef transform(data, label):\n return (nd.floor(data/128)).astype(np.float32), label.astype(np.float32)\nmnist_train = mx.gluon.data.vision.MNIST(train=True, transform=transform)\nmnist_test = mx.gluon.data.vision.MNIST(train=False, transform=transform)\n\n# Initialize the count statistics for p(y) and p(x_i|y)\n# We initialize all numbers with a count of 1 to ensure that we don't get a\n# division by zero. Statisticians call this Laplace smoothing.\nycount = nd.ones(shape=(10))\nxcount = nd.ones(shape=(784, 10))\n\n# Aggregate count statistics of how frequently a pixel is on (or off) for\n# zeros and ones.\nfor data, label in mnist_train:\n x = data.reshape((784,))\n y = int(label)\n ycount[y] += 1\n xcount[:, y] += x\n\n# normalize the probabilities p(x_i|y) (divide per pixel counts by total\n# count)\nfor i in range(10):\n xcount[:, i] = xcount[:, i]/ycount[i]\n\n# likewise, compute the probability p(y)\npy = ycount / nd.sum(ycount)", "_____no_output_____" ] ], [ [ "Now that we computed per-pixel counts of occurrence for all pixels, it's time to see how our model behaves. Time to plot it. We show the estimated probabilities of observing a switched-on pixel. These are some mean looking digits.", "_____no_output_____" ] ], [ [ "import matplotlib.pyplot as plt\nfig, figarr = plt.subplots(1, 10, figsize=(15, 15))\nfor i in range(10):\n figarr[i].imshow(xcount[:, i].reshape((28, 28)).asnumpy(), cmap='hot')\n figarr[i].axes.get_xaxis().set_visible(False)\n figarr[i].axes.get_yaxis().set_visible(False)\n\nplt.show()\nprint(py)", "_____no_output_____" ] ], [ [ "Now we can compute the likelihoods of an image, given the model. This is statistican speak for $p(x|y)$, i.e. how likely it is to see a particular image under certain conditions (such as the label). Since this is computationally awkward (we might have to multiply many small numbers if many pixels have a small probability of occurring), we are better off computing its logarithm instead. That is, instead of $p(x|y) = \\prod_{i} p(x_i|y)$ we compute $\\log p(x|y) = \\sum_i \\log p(x_i|y)$. \n\n$$l_y := \\sum_i \\log p(x_i|y) = \\sum_i x_i \\log p(x_i = 1|y) + (1-x_i) \\log \\left(1-p(x_i=1|y)\\right)$$\n\nTo avoid recomputing logarithms all the time, we precompute them for all pixels. ", "_____no_output_____" ] ], [ [ "logxcount = nd.log(xcount)\nlogxcountneg = nd.log(1-xcount)\nlogpy = nd.log(py)\n\nfig, figarr = plt.subplots(2, 10, figsize=(15, 3))\n\n# show 10 images\nctr = 0\nfor data, label in mnist_test:\n x = data.reshape((784,))\n y = int(label)\n \n # we need to incorporate the prior probability p(y) since p(y|x) is\n # proportional to p(x|y) p(y)\n logpx = logpy.copy()\n for i in range(10):\n # compute the log probability for a digit\n logpx[i] += nd.dot(logxcount[:, i], x) + nd.dot(logxcountneg[:, i], 1-x)\n # normalize to prevent overflow or underflow by subtracting the largest\n # value\n logpx -= nd.max(logpx)\n # and compute the softmax using logpx\n px = nd.exp(logpx).asnumpy()\n px /= np.sum(px)\n\n # bar chart and image of digit\n figarr[1, ctr].bar(range(10), px)\n figarr[1, ctr].axes.get_yaxis().set_visible(False)\n figarr[0, ctr].imshow(x.reshape((28, 28)).asnumpy(), cmap='hot')\n figarr[0, ctr].axes.get_xaxis().set_visible(False)\n figarr[0, ctr].axes.get_yaxis().set_visible(False)\n ctr += 1\n if ctr == 10:\n break\n\nplt.show()", "_____no_output_____" ] ], [ [ "As we can see, this classifier is both incompetent and overly confident of its incorrect estimates. That is, even if it is horribly wrong, it generates probabilities close to 1 or 0. Not a classifier we should use very much nowadays any longer. While Naive Bayes classifiers used to be popular in the 80s and 90s, e.g. for spam filtering, their heydays are over. The poor performance is due to the incorrect statistical assumptions that we made in our model: we assumed that each and every pixel are *independently* generated, depending only on the label. This is clearly not how humans write digits, and this wrong assumption led to the downfall of our overly naive (Bayes) classifier.", "_____no_output_____" ], [ "## Sampling\n\nRandom numbers are just one form of random variables, and since computers are particularly good with numbers, pretty much everything else in code ultimately gets converted to numbers anyway. One of the basic tools needed to generate random numbers is to sample from a distribution. Let's start with what happens when we use a random number generator. ", "_____no_output_____" ] ], [ [ "import random\nfor i in range(10):\n print(random.random())", "0.970844720223\n0.11442244666\n0.476145849846\n0.154138063676\n0.925771401913\n0.347466944833\n0.288795056587\n0.855051122608\n0.32666729925\n0.932922304219\n" ] ], [ [ "### Uniform Distribution\n\nThese are some pretty random numbers. As we can see, their range is between 0 and 1, and they are evenly distributed. That is, there is (actually, should be, since this is not a *real* random number generator) no interval in which numbers are more likely than in any other. In other words, the chances of any of these numbers to fall into the interval, say $[0.2,0.3)$ are as high as in the interval $[.593264, .693264)$. The way they are generated internally is to produce a random integer first, and then divide it by its maximum range. If we want to have integers directly, try the following instead. It generates random numbers between 0 and 100.", "_____no_output_____" ] ], [ [ "for i in range(10):\n print(random.randint(1, 100))", "75\n23\n34\n85\n99\n66\n13\n42\n19\n14\n" ] ], [ [ "What if we wanted to check that ``randint`` is actually really uniform. Intuitively the best strategy would be to run it, say 1 million times, count how many times it generates each one of the values and to ensure that the result is uniform. ", "_____no_output_____" ] ], [ [ "import math\n\ncounts = np.zeros(100)\nfig, axes = plt.subplots(2, 3, figsize=(15, 8), sharex=True)\naxes = axes.reshape(6)\n# mangle subplots such that we can index them in a linear fashion rather than\n# a 2d grid\n\nfor i in range(1, 1000001):\n counts[random.randint(0, 99)] += 1\n if i in [10, 100, 1000, 10000, 100000, 1000000]:\n axes[int(math.log10(i))-1].bar(np.arange(1, 101), counts)\nplt.show()", "_____no_output_____" ] ], [ [ "What we can see from the above figures is that the initial number of counts looks *very* uneven. If we sample fewer than 100 draws from a distribution over 100 outcomes this is pretty much expected. But even for 1000 samples there is a significant variability between the draws. What we are really aiming for is a situation where the probability of drawing a number $x$ is given by $p(x)$. \n\n### The categorical distribution\n\nQuite obviously, drawing from a uniform distribution over a set of 100 outcomes is quite simple. But what if we have nonuniform probabilities? Let's start with a simple case, a biased coin which comes up heads with probability 0.35 and tails with probability 0.65. A simple way to sample from that is to generate a uniform random variable over $[0,1]$ and if the number is less than $0.35$, we output heads and otherwise we generate tails. Let's try this out.", "_____no_output_____" ] ], [ [ "# number of samples\nn = 1000000\ny = np.random.uniform(0, 1, n)\nx = np.arange(1, n+1)\n# count number of occurrences and divide by the number of total draws\np0 = np.cumsum(y < 0.35) / x\np1 = np.cumsum(y >= 0.35) / x\n\nplt.figure(figsize=(15, 8))\nplt.semilogx(x, p0)\nplt.semilogx(x, p1)\nplt.show()", "_____no_output_____" ] ], [ [ "As we can see, on average this sampler will generate 35% zeros and 65% ones. Now what if we have more than two possible outcomes? We can simply generalize this idea as follows. Given any probability distribution, e.g. \n$p = [0.1, 0.2, 0.05, 0.3, 0.25, 0.1]$ we can compute its cumulative distribution (python's ``cumsum`` will do this for you) $F = [0.1, 0.3, 0.35, 0.65, 0.9, 1]$. Once we have this we draw a random variable $x$ from the uniform distribution $U[0,1]$ and then find the interval where $F[i-1] \\leq x < F[i]$. We then return $i$ as the sample. By construction, the chances of hitting interval $[F[i-1], F[i])$ has probability $p(i)$. \n\nNote that there are many more efficient algorithms for sampling than the one above. For instance, binary search over $F$ will run in $O(\\log n)$ time for $n$ random variables. There are even more clever algorithms, such as the [Alias Method](https://en.wikipedia.org/wiki/Alias_method) to sample in constant time, after $O(n)$ preprocessing.", "_____no_output_____" ], [ "### The Normal distribution\n\nThe Normal distribution (aka the Gaussian distribution) is given by $p(x) = \\frac{1}{\\sqrt{2 \\pi}} \\exp\\left(-\\frac{1}{2} x^2\\right)$. Let's plot it to get a feel for it.", "_____no_output_____" ] ], [ [ "x = np.arange(-10, 10, 0.01)\np = (1/math.sqrt(2 * math.pi)) * np.exp(-0.5 * x**2)\nplt.figure(figsize=(10, 5))\nplt.plot(x, p)\nplt.show()", "_____no_output_____" ] ], [ [ "Sampling from this distribution is a lot less trivial. First off, the support is infinite, that is, for any $x$ the density $p(x)$ is positive. Secondly, the density is nonuniform. There are many tricks for sampling from it - the key idea in all algorithms is to stratify $p(x)$ in such a way as to map it to the uniform distribution $U[0,1]$. One way to do this is with the probability integral transform. \n\nDenote by $F(x) = \\int_{-\\infty}^x p(z) dz$ the cumulative distribution function (CDF) of $p$. This is in a way the continuous version of the cumulative sum that we used previously. In the same way we can now define the inverse map $F^{-1}(\\xi)$, where $\\xi$ is drawn uniformly. Unlike previously where we needed to find the correct interval for the vector $F$ (i.e. for the piecewise constant function), we now invert the function $F(x)$. \n\nIn practice, this is slightly more tricky since inverting the CDF is hard in the case of a Gaussian. It turns out that the *twodimensional* integral is much easier to deal with, thus yielding two normal random variables than one, albeit at the price of two uniformly distributed ones. For now, suffice it to say that there are built-in algorithms to address this. \n\nThe normal distribution has yet another desirable property. In a way all distributions converge to it, if we only average over a sufficiently large number of draws from any other distribution. To understand this in a bit more detail, we need to introduce three important things: expected values, means and variances. \n\n* The expected value $\\mathbb{E}_{x \\sim p(x)}[f(x)]$ of a function $f$ under a distribution $p$ is given by the integral $\\int_x p(x) f(x) dx$. That is, we average over all possible outcomes, as given by $p$. \n* A particularly important expected value is that for the function $f(x) = x$, i.e. $\\mu := \\mathbb{E}_{x \\sim p(x)}[x]$. It provides us with some idea about the typical values of $x$.\n* Another important quantity is the variance, i.e. the typical deviation from the mean \n$\\sigma^2 := \\mathbb{E}_{x \\sim p(x)}[(x-\\mu)^2]$. Simple math shows (check it as an exercise) that\n$\\sigma^2 = \\mathbb{E}_{x \\sim p(x)}[x^2] - \\mathbb{E}^2_{x \\sim p(x)}[x]$.\n\nThe above allows us to change both mean and variance of random variables. Quite obviously for some random variable $x$ with mean $\\mu$, the random variable $x + c$ has mean $\\mu + c$. Moreover, $\\gamma x$ has the variance $\\gamma^2 \\sigma^2$. Applying this to the normal distribution we see that one with mean $\\mu$ and variance $\\sigma^2$ has the form $p(x) = \\frac{1}{\\sqrt{2 \\sigma^2 \\pi}} \\exp\\left(-\\frac{1}{2 \\sigma^2} (x-\\mu)^2\\right)$. Note the scaling factor $\\frac{1}{\\sigma}$ - it arises from the fact that if we stretch the distribution by $\\sigma$, we need to lower it by $\\frac{1}{\\sigma}$ to retain the same probability mass (i.e. the weight under the distribution always needs to integrate out to 1). \n\nNow we are ready to state one of the most fundamental theorems in statistics, the [Central Limit Theorem](https://en.wikipedia.org/wiki/Central_limit_theorem). It states that for sufficiently well-behaved random variables, in particular random variables with well-defined mean and variance, the sum tends toward a normal distribution. To get some idea, let's repeat the experiment described in the beginning, but now using random variables with integer values of $\\{0, 1, 2\\}$. ", "_____no_output_____" ] ], [ [ "# generate 10 random sequences of 10,000 random normal variables N(0,1)\ntmp = np.random.uniform(size=(10000,10))\nx = 1.0 * (tmp > 0.3) + 1.0 * (tmp > 0.8)\nmean = 1 * 0.5 + 2 * 0.2\nvariance = 1 * 0.5 + 4 * 0.2 - mean**2\nprint('mean {}, variance {}'.format(mean, variance))\n# cumulative sum and normalization\ny = np.arange(1,10001).reshape(10000,1)\nz = np.cumsum(x,axis=0) / y\n\nplt.figure(figsize=(10,5))\nfor i in range(10):\n plt.semilogx(y,z[:,i])\n\nplt.semilogx(y,(variance**0.5) * np.power(y,-0.5) + mean,'r')\nplt.semilogx(y,-(variance**0.5) * np.power(y,-0.5) + mean,'r')\nplt.show() ", "mean 0.9, variance 0.49\n" ] ], [ [ "This looks very similar to the initial example, at least in the limit of averages of large numbers of variables. This is confirmed by theory. Denote by mean and variance of a random variable the quantities \n\n$$\\mu[p] := \\mathbf{E}_{x \\sim p(x)}[x] \\text{ and } \\sigma^2[p] := \\mathbf{E}_{x \\sim p(x)}[(x - \\mu[p])^2]$$\n\nThen we have that $\\lim_{n\\to \\infty} \\frac{1}{\\sqrt{n}} \\sum_{i=1}^n \\frac{x_i - \\mu}{\\sigma} \\to \\mathcal{N}(0, 1)$. In other words, regardless of what we started out with, we will always converge to a Gaussian. This is one of the reasons why Gaussians are so popular in statistics.\n\n\n### More distributions\n\nMany more useful distributions exist. We recommend consulting a statistics book or looking some of them up on Wikipedia for further detail. \n\n* **Binomial Distribution** It is used to describe the distribution over multiple draws from the same distribution, e.g. the number of heads when tossing a biased coin (i.e. a coin with probability $\\pi$ of returning heads) 10 times. The probability is given by $p(x) = {n \\choose x} \\pi^x (1-\\pi)^{n-x}$. \n* **Multinomial Distribution** Obviously we can have more than two outcomes, e.g. when rolling a dice multiple times. In this case the distribution is given by $p(x) = \\frac{n!}{\\prod_{i=1}^k x_i!} \\prod_{i=1}^k \\pi_i^{x_i}$.\n* **Poisson Distribution** It is used to model the occurrence of point events that happen with a given rate, e.g. the number of raindrops arriving within a given amount of time in an area (weird fact - the number of Prussian soldiers being killed by horses kicking them followed that distribution). Given a rate $\\lambda$, the number of occurrences is given by $p(x) = \\frac{1}{x!} \\lambda^x e^{-\\lambda}$.\n* **Beta, Dirichlet, Gamma, and Wishart Distributions** They are what statisticians call *conjugate* to the Binomial, Multinomial, Poisson and Gaussian respectively. Without going into detail, these distributions are often used as priors for coefficients of the latter set of distributions, e.g. a Beta distribution as a prior for modeling the probability for binomial outcomes. ", "_____no_output_____" ], [ "## Next\n[Autograd](../chapter01_crashcourse/autograd.ipynb)", "_____no_output_____" ], [ "For whinges or inquiries, [open an issue on GitHub.](https://github.com/zackchase/mxnet-the-straight-dope)", "_____no_output_____" ] ] ]

[ "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown" ]

[ [ "markdown", "markdown" ], [ "code" ], [ "markdown" ], [ "code" ], [ "markdown" ], [ "code" ], [ "markdown" ], [ "code" ], [ "markdown" ], [ "code" ], [ "markdown" ], [ "code" ], [ "markdown" ], [ "code" ], [ "markdown" ], [ "code" ], [ "markdown" ], [ "code" ], [ "markdown", "markdown", "markdown", "markdown" ], [ "code" ], [ "markdown" ], [ "code" ], [ "markdown" ], [ "code" ], [ "markdown", "markdown" ], [ "code" ], [ "markdown" ], [ "code" ], [ "markdown" ], [ "code" ], [ "markdown" ], [ "code" ], [ "markdown", "markdown" ], [ "code" ], [ "markdown" ], [ "code" ], [ "markdown", "markdown", "markdown" ] ]

[ "Apache-2.0" ]

[ "Apache-2.0" ]

[ "Apache-2.0" ]

[ [ [ "___\n\n<a href='https://www.udemy.com/user/joseportilla/'><img src='../Pierian_Data_Logo.png'/></a>\n___\n<center><em>Content Copyright by Pierian Data</em></center>", "_____no_output_____" ], [ "# Statements Assessment Solutions", "_____no_output_____" ], [ "_____\n**Use <code>for</code>, .split(), and <code>if</code> to create a Statement that will print out words that start with 's':**", "_____no_output_____" ] ], [ [ "st = 'Print only the words that start with s in this sentence'", "_____no_output_____" ], [ "for word in st.split():\n if word[0] == 's':\n print(word)", "start\ns\nsentence\n" ] ], [ [ "______\n**Use range() to print all the even numbers from 0 to 10.**", "_____no_output_____" ] ], [ [ "list(range(0,11,2))", "_____no_output_____" ] ], [ [ "___\n**Use List comprehension to create a list of all numbers between 1 and 50 that are divisible by 3.**", "_____no_output_____" ] ], [ [ "[x for x in range(1,51) if x%3 == 0]", "_____no_output_____" ] ], [ [ "_____\n**Go through the string below and if the length of a word is even print \"even!\"**", "_____no_output_____" ] ], [ [ "st = 'Print every word in this sentence that has an even number of letters'", "_____no_output_____" ], [ "for word in st.split():\n if len(word)%2 == 0:\n print(word+\" <-- has an even length!\")", "word <-- has an even length!\nin <-- has an even length!\nthis <-- has an even length!\nsentence <-- has an even length!\nthat <-- has an even length!\nan <-- has an even length!\neven <-- has an even length!\nnumber <-- has an even length!\nof <-- has an even length!\n" ] ], [ [ "____\n**Write a program that prints the integers from 1 to 100. But for multiples of three print \"Fizz\" instead of the number, and for the multiples of five print \"Buzz\". For numbers which are multiples of both three and five print \"FizzBuzz\".**", "_____no_output_____" ] ], [ [ "for num in range(1,101):\n if num % 3 == 0 and num % 5 == 0:\n print(\"FizzBuzz\")\n elif num % 3 == 0:\n print(\"Fizz\")\n elif num % 5 == 0:\n print(\"Buzz\")\n else:\n print(num)", "_____no_output_____" ] ], [ [ "____\n**Use a List Comprehension to create a list of the first letters of every word in the string below:**", "_____no_output_____" ] ], [ [ "st = 'Create a list of the first letters of every word in this string'", "_____no_output_____" ], [ "[word[0] for word in st.split()]", "_____no_output_____" ] ], [ [ "### Great Job!", "_____no_output_____" ] ] ]

[ "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown" ]

[ [ "markdown", "markdown", "markdown" ], [ "code", "code" ], [ "markdown" ], [ "code" ], [ "markdown" ], [ "code" ], [ "markdown" ], [ "code", "code" ], [ "markdown" ], [ "code" ], [ "markdown" ], [ "code", "code" ], [ "markdown" ] ]

[ "MIT" ]

[ "MIT" ]

[ "MIT" ]

[ [ [ "import numpy as np\nimport pandas as pd\nimport torch\nimport torch.nn as nn\nimport torch.nn.parallel\nimport torch.optim as optim\nimport torch.utils.data\nfrom torch.autograd import Variable\n\nmovies = pd.read_csv('mn/movies.dat',sep = '::', header =None, engine = 'python', encoding = 'latin-1')\n\nmovies\n\nusers = pd.read_csv('mn/users.dat',sep = '::', header =None, engine = 'python', encoding = 'latin-1')\n\nusers\n\nratings = pd.read_csv('mn/ratings.dat',sep = '::', header =None, engine = 'python', encoding = 'latin-1')\n\nratings\n\n#Preparing the Training set and Test set\n\ntraining_set = pd.read_csv('boltz/u1.base',delimiter ='\\t')\n\ntraining_set\n\ntraining_set = np.array(training_set, dtype = 'int')\n\n\ntest_set = pd.read_csv('boltz/u1.test',delimiter ='\\t')\ntest_set = np.array(training_set, dtype = 'int')\n\n#Getting the number of users and movies\n\nnb_users = int(max(max(training_set[:,0]),max(test_set[:,0]))) \nnb_movies =int(max(max(training_set[:,1]),max(test_set[:,1])))\n\nnb_users\n\nnb_movies\n\n#Convert data into array with users in lines and movies in columns\n\ndef convert(data):\n new_data = []\n for id_users in range(1,nb_users+1):\n id_movies = data[:, 1][data[:,0]==id_users]\n id_ratings = data[:, 2][data[:,0]==id_users]\n ratings = np.zeros(nb_movies)\n ratings[id_movies-1] = id_ratings\n new_data.append(list(ratings))\n return new_data\ntraining_set = convert(training_set)\ntest_set = convert(test_set)\n \n\ntraining_set\n\n#Converting into torch\n\ntraining_set = torch.FloatTensor(training_set)\ntest_set = torch.FloatTensor(test_set)\n\n#Creating the architecture of the Neural Network\n\nclass SAE(nn.Module):\n def __init(self, ):\n super(SAE, self).__init__()\n self.fc1 = nn.Linear(nb_movies, 20)\n self.fc2 = nn.Linear(20, 10)\n self.fc3 = nn.Linear(10,20)\n self.fc4 = nn.Linear(20,nb_movies)\n self.activation = nn.Sigmoid()\n def forward(self, x):\n x = self.activation(self.fc1(x))\n x = self.activation(self.fc2(x))\n x = self.activation(self.fc3(x))\n x = self.fc4(x)\n return x\nsae = SAE()\n\n", "_____no_output_____" ] ] ]

[ "code" ]

[ [ "code" ] ]

[ "MIT" ]

[ "MIT" ]

[ "MIT" ]

[ [ [ "### This is Example 4.3. Gambler’s Problem from Sutton's book.\n\nA gambler has the opportunity to make bets on the outcomes of a sequence of coin flips. \nIf the coin comes up heads, he wins as many dollars as he has staked on that flip; \nif it is tails, he loses his stake. The game ends when the gambler wins by reaching his goal of $100, \nor loses by running out of money. \n\nOn each flip, the gambler must decide what portion of his capital to stake, in integer numbers of dollars. \nThis problem can be formulated as an undiscounted, episodic, finite MDP. \n\nThe state is the gambler’s capital, s ∈ {1, 2, . . . , 99}.\nThe actions are stakes, a ∈ {0, 1, . . . , min(s, 100 − s)}. \nThe reward is zero on all transitions except those on which the gambler reaches his goal, when it is +1.\n\nThe state-value function then gives the probability of winning from each state. A policy is a mapping from levels of capital to stakes. The optimal policy maximizes the probability of reaching the goal. Let p_h denote the probability of the coin coming up heads. If p_h is known, then the entire problem is known and it can be solved, for instance, by value iteration.\n", "_____no_output_____" ] ], [ [ "import numpy as np\nimport sys\nimport matplotlib.pyplot as plt\nif \"../\" not in sys.path:\n sys.path.append(\"../\") ", "_____no_output_____" ] ], [ [ "\n### Exercise 4.9 (programming)\n\nImplement value iteration for the gambler’s problem and solve it for p_h = 0.25 and p_h = 0.55.", "_____no_output_____" ] ], [ [ "def value_iteration_for_gamblers(p_h, theta=0.0001, discount_factor=1.0):\n \"\"\"\n Args:\n p_h: Probability of the coin coming up heads\n \"\"\"\n # The reward is zero on all transitions except those on which the gambler reaches his goal,\n # when it is +1.\n rewards = np.zeros(101)\n rewards[100] = 1 \n \n # We introduce two dummy states corresponding to termination with capital of 0 and 100\n V = np.zeros(101)\n \n def one_step_lookahead(s, V, rewards):\n \"\"\"\n Helper function to calculate the value for all action in a given state.\n \n Args:\n s: The gambler’s capital. Integer.\n V: The vector that contains values at each state. \n rewards: The reward vector.\n \n Returns:\n A vector containing the expected value of each action. \n Its length equals to the number of actions.\n \"\"\"\n A = np.zeros(101)\n stakes = range(1, min(s, 100-s)+1) # Your minimum bet is 1, maximum bet is min(s, 100-s).\n for a in stakes:\n # rewards[s+a], rewards[s-a] are immediate rewards.\n # V[s+a], V[s-a] are values of the next states.\n # This is the core of the Bellman equation: The expected value of your action is \n # the sum of immediate rewards and the value of the next state.\n A[a] = p_h * (rewards[s+a] + V[s+a]*discount_factor) + (1-p_h) * (rewards[s-a] + V[s-a]*discount_factor)\n return A\n \n while True:\n # Stopping condition\n delta = 0\n # Update each state...\n for s in range(1, 100):\n # Do a one-step lookahead to find the best action\n A = one_step_lookahead(s, V, rewards)\n # print(s,A,V) # if you want to debug.\n best_action_value = np.max(A)\n # Calculate delta across all states seen so far\n delta = max(delta, np.abs(best_action_value - V[s]))\n # Update the value function. Ref: Sutton book eq. 4.10. \n V[s] = best_action_value \n # Check if we can stop \n if delta < theta:\n break\n \n # Create a deterministic policy using the optimal value function\n policy = np.zeros(100)\n for s in range(1, 100):\n # One step lookahead to find the best action for this state\n A = one_step_lookahead(s, V, rewards)\n best_action = np.argmax(A)\n # Always take the best action\n policy[s] = best_action\n \n return policy, V", "_____no_output_____" ], [ "policy, v = value_iteration_for_gamblers(0.25)\n\nprint(\"Optimized Policy:\")\nprint(policy)\nprint(\"\")\n\nprint(\"Optimized Value Function:\")\nprint(v)\nprint(\"\")", "Optimized Policy:\n[ 0. 1. 2. 3. 4. 5. 6. 7. 8. 9. 10. 11. 12. 12. 11. 15. 16. 17.\n 18. 6. 20. 21. 3. 23. 24. 25. 1. 2. 3. 4. 5. 6. 7. 8. 9. 10.\n 11. 12. 38. 11. 10. 9. 42. 7. 44. 5. 46. 47. 48. 49. 50. 1. 2. 3.\n 4. 5. 6. 7. 8. 9. 10. 11. 12. 13. 11. 10. 9. 17. 7. 19. 5. 21.\n 22. 23. 24. 25. 1. 2. 3. 4. 5. 6. 7. 8. 9. 10. 11. 12. 12. 11.\n 10. 9. 8. 7. 6. 5. 4. 3. 2. 1.]\n\nOptimized Value Function:\n[0.00000000e+00 7.24792480e-05 2.89916992e-04 6.95257448e-04\n 1.16010383e-03 1.76906586e-03 2.78102979e-03 4.03504074e-03\n 4.66214120e-03 5.59997559e-03 7.08471239e-03 9.03964043e-03\n 1.11241192e-02 1.56793594e-02 1.61464431e-02 1.69517994e-02\n 1.86512806e-02 1.98249817e-02 2.24047303e-02 2.73845196e-02\n 2.83388495e-02 3.04937363e-02 3.61633897e-02 3.84953022e-02\n 4.44964767e-02 6.25000000e-02 6.27174377e-02 6.33700779e-02\n 6.45857723e-02 6.59966059e-02 6.78135343e-02 7.08430894e-02\n 7.46098323e-02 7.64884604e-02 7.93035477e-02 8.37541372e-02\n 8.96225423e-02 9.58723575e-02 1.09538078e-01 1.10939329e-01\n 1.13360151e-01 1.18457374e-01 1.21977661e-01 1.29716907e-01\n 1.44653559e-01 1.47520113e-01 1.53983246e-01 1.70990169e-01\n 1.77987434e-01 1.95990576e-01 2.50000000e-01 2.50217438e-01\n 2.50870078e-01 2.52085772e-01 2.53496606e-01 2.55313534e-01\n 2.58343089e-01 2.62109832e-01 2.63988460e-01 2.66803548e-01\n 2.71254137e-01 2.77122542e-01 2.83372357e-01 2.97038078e-01\n 2.98439329e-01 3.00860151e-01 3.05957374e-01 3.09477661e-01\n 3.17216907e-01 3.32153559e-01 3.35020113e-01 3.41483246e-01\n 3.58490169e-01 3.65487434e-01 3.83490576e-01 4.37500000e-01\n 4.38152558e-01 4.40122454e-01 4.43757317e-01 4.47991345e-01\n 4.53440603e-01 4.62529268e-01 4.73829497e-01 4.79468031e-01\n 4.87912680e-01 5.01265085e-01 5.18867627e-01 5.37617932e-01\n 5.78614419e-01 5.82817988e-01 5.90080452e-01 6.05372123e-01\n 6.15934510e-01 6.39150720e-01 6.83960814e-01 6.92560339e-01\n 7.11950883e-01 7.62970611e-01 7.83963162e-01 8.37972371e-01\n 0.00000000e+00]\n\n" ] ], [ [ "### Show your results graphically, as in Figure 4.3.\n", "_____no_output_____" ] ], [ [ "# Plotting Final Policy (action stake) vs State (Capital)\n\n# x axis values\nx = range(100)\n# corresponding y axis values\ny = v[:100]\n \n# plotting the points \nplt.plot(x, y)\n \n# naming the x axis\nplt.xlabel('Capital')\n# naming the y axis\nplt.ylabel('Value Estimates')\n \n# giving a title to the graph\nplt.title('Final Policy (action stake) vs State (Capital)')\n \n# function to show the plot\nplt.show()", "_____no_output_____" ], [ "# Plotting Capital vs Final Policy\n\n# x axis values\nx = range(100)\n# corresponding y axis values\ny = policy\n \n# plotting the bars\nplt.bar(x, y, align='center', alpha=0.5)\n \n# naming the x axis\nplt.xlabel('Capital')\n# naming the y axis\nplt.ylabel('Final policy (stake)')\n \n# giving a title to the graph\nplt.title('Capital vs Final Policy')\n \n# function to show the plot\nplt.show()\n", "_____no_output_____" ] ] ]

[ "markdown", "code", "markdown", "code", "markdown", "code" ]

[ [ "markdown" ], [ "code" ], [ "markdown" ], [ "code", "code" ], [ "markdown" ], [ "code", "code" ] ]

[ "Apache-2.0" ]

[ "Apache-2.0" ]

[ "Apache-2.0" ]

[ [ [ "Amazon web scraper", "_____no_output_____" ] ], [ [ "import csv\nfrom bs4 import BeautifulSoup", "_____no_output_____" ], [ "# firefox and Chrome\nfrom selenium import webdriver", "_____no_output_____" ] ], [ [ " Startup the webdriver", "_____no_output_____" ] ], [ [ "pip install webdriver-manager", "Requirement already satisfied: webdriver-manager in c:\\users\\bikem\\anaconda3\\envs\\selenium_3.8\\lib\\site-packages (3.5.2)\nRequirement already satisfied: requests in c:\\users\\bikem\\anaconda3\\envs\\selenium_3.8\\lib\\site-packages (from webdriver-manager) (2.27.1)\nRequirement already satisfied: configparser in c:\\users\\bikem\\anaconda3\\envs\\selenium_3.8\\lib\\site-packages (from webdriver-manager) (5.2.0)\nRequirement already satisfied: crayons in c:\\users\\bikem\\anaconda3\\envs\\selenium_3.8\\lib\\site-packages (from webdriver-manager) (0.4.0)Note: you may need to restart the kernel to use updated packages.\n\nRequirement already satisfied: colorama in c:\\users\\bikem\\anaconda3\\envs\\selenium_3.8\\lib\\site-packages (from crayons->webdriver-manager) (0.4.4)\nRequirement already satisfied: certifi>=2017.4.17 in c:\\users\\bikem\\anaconda3\\envs\\selenium_3.8\\lib\\site-packages (from requests->webdriver-manager) (2021.10.8)\nRequirement already satisfied: urllib3<1.27,>=1.21.1 in c:\\users\\bikem\\anaconda3\\envs\\selenium_3.8\\lib\\site-packages (from requests->webdriver-manager) (1.26.7)\nRequirement already satisfied: charset-normalizer~=2.0.0 in c:\\users\\bikem\\anaconda3\\envs\\selenium_3.8\\lib\\site-packages (from requests->webdriver-manager) (2.0.4)\nRequirement already satisfied: idna<4,>=2.5 in c:\\users\\bikem\\anaconda3\\envs\\selenium_3.8\\lib\\site-packages (from requests->webdriver-manager) (3.3)\n" ], [ "## we activate the webdriver for Chrome as we are using Google Chrome\nfrom webdriver_manager.chrome import ChromeDriverManager\n\ndriver = webdriver.Chrome(ChromeDriverManager().install())", "\n\n====== WebDriver manager ======\nCurrent google-chrome version is 97.0.4692\nGet LATEST chromedriver version for 97.0.4692 google-chrome\nDriver [C:\\Users\\bikem\\.wdm\\drivers\\chromedriver\\win32\\97.0.4692.71\\chromedriver.exe] found in cache\nC:\\Users\\bikem\\AppData\\Local\\Temp/ipykernel_16024/3985819463.py:4: DeprecationWarning: executable_path has been deprecated, please pass in a Service object\n driver = webdriver.Chrome(ChromeDriverManager().install())\n" ], [ "# Using webdriver we'll now open the Amazon website in chrome\nurl = 'https://www.amazon.in'\n\n# We'll use the get method of driver and pass in the URL\ndriver.get(url)", "_____no_output_____" ], [ "def get_url(search_term):\n '''\n This function fetches the URL of the item that you want to search\n '''\n template = 'https://www.amazon.in/s?k={}&crid=UOAX8JTJZ8XJ&ref=nb_sb_noss_1'\n # We'are replacing every space with '+' to adhere with the pattern \n search_term = search_term.replace(\" \",\"+\")\n return template.format(search_term)", "_____no_output_____" ], [ "# Checking whether the function is working properly or not\nurl = get_url('nike shoes')\nprint(url)", "https://www.amazon.in/s?k=nike+shoes&crid=UOAX8JTJZ8XJ&ref=nb_sb_noss_1\n" ], [ "driver.get(url)", "_____no_output_____" ] ], [ [ "**Extract the collection**", "_____no_output_____" ] ], [ [ "#taking the page source and trying to extract from html\nsoup = BeautifulSoup(driver.page_source, 'html.parser')", "_____no_output_____" ], [ "# assigning the specific identity of the component we need to extract from the website\n# in this case we need to extract the whole component that we search in the site\n# say the mobile phone in this case and the whole component containing the name, price, etc needs to be assigned\nresults = soup.find_all('div', {'data-component-type': 's-search-result'})", "_____no_output_____" ], [ "len(results)", "_____no_output_____" ], [ "# prototype the results\nitem = results[0]", "_____no_output_____" ], [ "item.find('div', 'a-section a-spacing-medium a-text-center').text", "_____no_output_____" ], [ "''' while we try and extract the first most obvious thing to select would be the name of the product\n in order to select that we see that the name component was under the h2 tag and under a, hence \n we extract that and assign to atag'''\natag = item.h2.a", "_____no_output_____" ], [ "atag.text.strip()", "_____no_output_____" ], [ "# we select that text from the atag and strip to select only the name text component\ndescription = atag.text.strip()", "_____no_output_____" ], [ "# we need the exact url point of this component to be extracted properly\natag.get('href')", "_____no_output_____" ], [ "# additionally we need this to be prefixed with the https amazon tag\nurl = 'https://www.amazon.in' + atag.get('href')", "_____no_output_____" ], [ "# now we need to extract the price of the product\nitem.find('span', 'a-price-whole').text", "_____no_output_____" ], [ "# now we need to extract the price of the product\nprice = item.find('span', 'a-price-whole').text", "_____no_output_____" ], [ "item.i.text", "_____no_output_____" ], [ "rating = item.i.text", "_____no_output_____" ], [ "# to extract the number of reviews given to the product\nitem.find('span', 'a-size-base').text", "_____no_output_____" ], [ "rating_count = item.find('span', 'a-size-base').text", "_____no_output_____" ] ], [ [ "## Generalize the pattern now", "_____no_output_____" ] ], [ [ "def extract_records(item):\n '''Extract and return data from a single record'''\n #description and url\n atag = item.h2.a\n description = atag.text.strip()\n url = \"https://www.amazon.in\" + atag.get(\"href\")\n \n # price\n price = item.find('span', 'a-price-whole').text\n \n # rank and rating\n rating = item.i.text\n \n item.find('span', 'a-size-base').text\n rating_count = item.find('span', 'a-size-base').text\n \n result = (description, price, rating, rating_count, url)\n \n return result", "_____no_output_____" ], [ "# now we try to apply the above to the url and try to extract\nrecords = []\nresults = soup.find_all('div', {'data-component-type': 's-search-result'})\n\nfor item in results:\n records.append(extract_records(item))", "_____no_output_____" ] ], [ [ "**We encounter the attribute error, this is basically because there are numerous results in the website page that need not match all the description as given for each of the products. Not all products might be having the same descriptions. Hence the attribute error occuring. We need to give exception for this error in the code**", "_____no_output_____" ], [ "## Error Handling", "_____no_output_____" ] ], [ [ "def extract_records(item):\n '''Extract and return data from a single record'''\n #description and url\n atag = item.h2.a\n description = atag.text.strip()\n url = \"https://www.amazon.in\" + atag.get(\"href\")\n \n '''Basically we put the exception for price here which most definitely \n should not be there return'''\n try:\n # price\n price = item.find('span', 'a-price-whole').text\n except AttributeError:\n price = ''\n \n \n ''' For the ratings although there might be missign details and \n we rather add the exception properly'''\n try:\n # rank and rating\n rating = item.i.text\n \n item.find('span', 'a-size-base').text\n rating_count = item.find('span', 'a-size-base').text\n except AttributeError:\n rating = ''\n rating_count = ''\n \n result = (description, price, rating, rating_count, url)\n \n return result", "_____no_output_____" ], [ "# additionally we need to check whether the results give any empty records\nrecords = []\nresults = soup.find_all('div', {'data-component-type': 's-search-result'})\n\nfor item in results:\n records.append(extract_records(item))", "_____no_output_____" ], [ "records[0]", "_____no_output_____" ], [ "for row in records:\n print(row[1])", "3,240\n\n3,763\n5,596\n1,560\n4,103\n5,598\n1,729\n2,580\n2,412\n5,072\n3,652\n3,614\n5,493\n6,293\n4,224\n1,399\n1,599\n1,599\n1,399\n3,101\n\n2,506\n3,294\n4,273\n2,464\n2,654\n3,694\n2,397\n3,287\n5,995\n2,747\n5,852\n3,596\n1,896\n2,700\n2,346\n2,995\n6,942\n2,694\n2,018\n3,795\n3,477\n2,036\n3,357\n2,408\n8,588\n3,326\n5,396\n5,260\n8,109\n3,179\n13,019\n6,999\n449\n1,399\n1,399\n1,499\n9,899\n1,599\n" ], [ "# next step is to get the next page in order to get most reviews\n# we need to take the component for the next button\ndef get_url(search_term):\n '''\n This function fetches the URL of the item that you want to search\n '''\n template = 'https://www.amazon.in/s?k={}&crid=UOAX8JTJZ8XJ&ref=nb_sb_noss_1'\n # We'are replacing every space with '+' to adhere with the pattern \n search_term = search_term.replace(\" \",\"+\")\n \n # add term query to url\n url = template.format(search_term)\n \n # add page query placeholder\n url += '&page{}'\n \n return url", "_____no_output_____" ], [ "## Putting all the pieces of code together\nimport csv\nfrom bs4 import BeautifulSoup\nfrom selenium import webdriver\nfrom webdriver_manager.chrome import ChromeDriverManager\n\ndef get_url(search_term):\n '''\n This function fetches the URL of the item that you want to search\n '''\n template = 'https://www.amazon.in/s?k={}&crid=UOAX8JTJZ8XJ&ref=nb_sb_noss_1'\n # We'are replacing every space with '+' to adhere with the pattern \n search_term = search_term.replace(\" \",\"+\")\n \n # add term query to url\n url = template.format(search_term)\n \n # add page query placeholder\n url += '&page{}'\n \n return url\n\ndef extract_records(item):\n '''Extract and return data from a single record'''\n #description and url\n atag = item.h2.a\n description = atag.text.strip()\n url = \"https://www.amazon.in\" + atag.get(\"href\")\n \n '''Basically we put the exception for price here which most definitely \n should not be there return'''\n try:\n # price\n price = item.find('span', 'a-price-whole').text\n except AttributeError:\n price = ''\n \n ''' For the ratings although there might be missign details and \n we rather add the exception properly'''\n try:\n # rank and rating\n rating = item.i.text\n \n item.find('span', 'a-size-base').text\n rating_count = item.find('span', 'a-size-base').text\n except AttributeError:\n rating = ''\n rating_count = ''\n \n result = (description, price, rating, rating_count, url)\n \n return result\n\ndef main(search_term):\n \"\"\"Run the main program routine\"\"\"\n driver = webdriver.Chrome(ChromeDriverManager().install())\n\n record = []\n url = get_url(search_term)\n \n for page in range(1, 6):\n driver.get(url.format(page))\n soup = BeautifulSoup(driver.page_source, 'html.parser')\n results = soup.find_all('div', {'data-component-type': 's-search-result'})\n \n for item in results:\n record = extract_records(item)\n if record:\n records.append(record)\n \n driver.close()\n \n #save the data to csv file\n with open('results.csv', 'w', newline='', encoding = 'utf-8') as f:\n writer = csv.writer(f)\n writer.writerow({'Description', 'Price_in_INR', 'Reviews', 'Review_count', 'Url'})\n writer.writerows(records)", "_____no_output_____" ], [ "main('nike shoes')", "\n\n====== WebDriver manager ======\nCurrent google-chrome version is 97.0.4692\nGet LATEST chromedriver version for 97.0.4692 google-chrome\nDriver [C:\\Users\\bikem\\.wdm\\drivers\\chromedriver\\win32\\97.0.4692.71\\chromedriver.exe] found in cache\nC:\\Users\\bikem\\AppData\\Local\\Temp/ipykernel_16024/2398463386.py:56: DeprecationWarning: executable_path has been deprecated, please pass in a Service object\n driver = webdriver.Chrome(ChromeDriverManager().install())\n" ] ] ]

[ "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code" ]

[ [ "markdown" ], [ "code", "code" ], [ "markdown" ], [ "code", "code", "code", "code", "code", "code" ], [ "markdown" ], [ "code", "code", "code", "code", "code", "code", "code", "code", "code", "code", "code", "code", "code", "code", "code", "code" ], [ "markdown" ], [ "code", "code" ], [ "markdown", "markdown" ], [ "code", "code", "code", "code", "code", "code", "code" ] ]

[ "MIT" ]

[ "MIT" ]

[ "MIT" ]

[ [ [ "# Test the hypothesis whether GRE/TOEFL grades play pivotal role in graduate admissions", "_____no_output_____" ] ], [ [ "# for some basic operations\nimport numpy as np\nimport pandas as pd\n\n# for data visualizations\nimport matplotlib.pyplot as plt\nimport seaborn as sns\nplt.style.use('fivethirtyeight')\n\n# for advanced visualizations\nimport plotly.offline as py\nfrom plotly.offline import init_notebook_mode, iplot\nimport plotly.graph_objs as go\nfrom plotly import tools\ninit_notebook_mode(connected = True)\nimport plotly.figure_factory as ff\n\n# for providing path\nimport os\nprint(os.listdir('/Users/sertan/Documents/DataScience_GT/TheProject/Bladder-Cancer-Detection/'))", "_____no_output_____" ], [ "# reading the data\n\ndata = pd.read_csv('/Users/sertan/Documents/DataScience_GT/TheProject/Bladder-Cancer-Detection/Admission_Predict_Ver1.1.csv')\n#data2 = pd.read_csv('/Users/sertan/Documents/DataScience_GT/TheProject/Bladder-Cancer-Detection/input/Admission_Predict.csv')\n\n# getting the shapes of the datasets\nprint(\"Shape of data1: \", data.shape)\n#print(\"Shape of data2 :\", data2.shape)\n\n# combining both the datasets as they have same columns\n\n#data = pd.concat([data, data2])\n\n# getting the shape of new dataset\ndata.head()", "Shape of data1: (500, 9)\n" ], [ "# checking if the data contains any NULL values\n\ndata.isnull().any().any()", "_____no_output_____" ], [ "data.describe()", "_____no_output_____" ], [ "# looking at the variations of LOR among the students\n\nplt.rcParams['figure.figsize'] = (18, 9)\nplt.style.use('dark_background')\n\nsns.countplot(data['LOR '], palette = 'PuBu')\nplt.title('Variations in Letter of Recommendations', fontsize = 30)\nplt.xlabel('LOR Score')\nplt.ylabel('count')\nplt.show()", "_____no_output_____" ], [ "# making a pie chart for the analysis of students rather they did research or not.\n\ndata_re = data['Research'].value_counts()\n\nlabel_re = data_re.index\nsize_re = data_re.values\n\ncolors = ['aqua', 'gold']\n\ntrace = go.Pie(\n labels = label_re, values = size_re, marker = dict(colors = colors), name = 'Research', hole = 0.3)\n\ndf = [trace]\n\nlayout1 = go.Layout(\n title = 'Research work done or not')\nfig = go.Figure(data = df, layout = layout1)\npy.iplot(fig)", "_____no_output_____" ], [ "# making a donut chart for the analysis of students with different university ratings\n\ndata_ur = data['University Rating'].value_counts()\n\nlabel_re = data_ur.index\nsize_re = data_ur.values\n\n\ntrace = go.Pie(\n labels = label_re,\n values = size_re,\n marker = dict(colors = ['gold' 'lightgreen', 'orange', 'yellow', 'pink']),\n name = 'University Ratings',\n hole = 0.2)\n\ndf2 = [trace]\n\nlayout1 = go.Layout(\n title = 'University Ratings of the Students')\nfig = go.Figure(data = df2, layout = layout1)\npy.iplot(fig)", "_____no_output_____" ], [ "trace = go.Box(\n x = data['University Rating'],\n y = data['GRE Score'],\n name = 'University Rating vs GRE Score',\n marker = dict(\n color = 'rgb(145, 165, 5)')\n)\n \n\ndf= [trace]\n\nlayout = go.Layout(\n boxmode = 'group',\n title = 'University Ratings vs GRE Score',\n \n)\n\nfig = go.Figure(data = df, layout = layout)\npy.iplot(fig)", "_____no_output_____" ], [ "plt.rcParams['figure.figsize'] = (18, 9)\nplt.style.use('ggplot')\n\nsns.boxenplot(data['University Rating'], data['TOEFL Score'], palette = 'RdPu')\nplt.title('University Ratings vs TOEFL Score', fontsize = 20)\nplt.show()", "_____no_output_____" ], [ "plt.rcParams['figure.figsize'] = (18, 9)\nplt.style.use('ggplot')\n\nsns.swarmplot(data['University Rating'], data['CGPA'], palette = 'twilight')\nplt.title('University Ratings vs CGPA', fontsize = 20)\nplt.show()", "_____no_output_____" ], [ "plt.rcParams['figure.figsize'] = (18, 9)\nplt.style.use('ggplot')\n\nsns.violinplot(data['University Rating'], data['Chance of Admit '], palette = 'rainbow')\nplt.title('University Ratings vs Chance of Admission', fontsize = 20)\nplt.show()", "_____no_output_____" ], [ "# prepare data\n\ndata2 = data.loc[:,[\"GRE Score\", \"TOEFL Score\", \"Chance of Admit \"]]\ndata2[\"index\"] = np.arange(1,len(data2)+1)\n\n# scatter matrix\nfig = ff.create_scatterplotmatrix(data2, diag='box', index='index',colormap='Portland',\n colormap_type='cat',\n height=700, width=700)\niplot(fig)", "_____no_output_____" ] ], [ [ "# Based on the correlation plots above, we concluded that GRE/TOEFL grades do play a pivotal role in graduate admissions", "_____no_output_____" ] ] ]

[ "markdown", "code", "markdown" ]

[ [ "markdown" ], [ "code", "code", "code", "code", "code", "code", "code", "code", "code", "code", "code", "code" ], [ "markdown" ] ]

[ "MIT" ]

[ "MIT" ]

[ "MIT" ]

[ [ [ "<img src='./img/intel-logo.jpg' width=30%, Fig1> \n\n# 파이썬 기초강의\n<font size=5><b>01. 연 습 문 제<b></font>\n\n<div align='right'>이 인 구 (Ike Lee)</div>\n", "_____no_output_____" ], [ "## Example : 1\n\n### 직육면체의 부피를 구해보자\n<img src='./img/volume.png' width=50%, Fig2>", "_____no_output_____" ] ], [ [ "# 변수 설정 \nlength = 5\nheight = 5\nwidth = 20", "_____no_output_____" ], [ "volume = length*width*height\nprint('직육면체의 부피 : %d'%volume)", "_____no_output_____" ], [ "length = 10 # 다시 할당하면 됨\nvolume = length*width*height\nprint('직육면체의 부피 : %d'%volume)", "_____no_output_____" ] ], [ [ "## Example : 2\n### for 를 사용해서 암컷 개를 찾으세요. and를 사용해서 일치하는것을 찾으세요 ", "_____no_output_____" ] ], [ [ "suspects = [['낙타', '포유류','암컷'], ['상어','어류','숫컷'], ['푸들','개','암컷']]\nfor suspect in suspects:\n if suspect[1] == '개' and suspect[2] =='암컷':\n print('범인은', suspect[0], '입니다')", "_____no_output_____" ], [ "volume = length*width*height\nprint('직육면체의 부피 : %d'%volume)", "_____no_output_____" ], [ "length = 10 # 다시 할당하면 됨\nvolume = length*width*height\nprint('직육면체의 부피 : %d'%volume)", "_____no_output_____" ] ], [ [ "## Example : 3\n### 연이율구하기 \n```\n 2017년 7월 2일 연이율 3% 계좌를 생성하여 3000000원을 입금한 경우\n 2018년 7월 2일 계좌 총액을 계산하여 출력하는 프로그램을 작성하십시오.\n 프로그램에서 입금액을 위한 변수, 연이율을 위한 변수를 만들어 사용하십시오.\n\n위의 프로그램을 입금액과 연이율을 입력받아 총액을 출력하도록 변경하십시오.\n언어 : python3\n입력 설명 :\n\n다음은 입금액과 연이율의 입력예입니다.\n===============================\n입금액(원), 연이율(%):: 4000, 3\n출력 설명 :\n\n다음과 같이 1년 후 총액을 출력합니다.\n===============================\n4120.0\n\n샘플 입력 : 4000, 3\n샘플 출력 : 4120.0\n``` ", "_____no_output_____" ] ], [ [ "money, ratio = eval(input('입금액(원), 연이율(%)::'))\nprint(money*(1+(1/100)*ratio))", "_____no_output_____" ] ], [ [ "## Example : 4\n### 삼각형 넓이를 구하시오\n```\n\n삼각형의 세변의 길이가 3,4,5인 경우 삼각형 넓이는 다음과 같이 계산합니다.\n x = (3 + 4 + 5)/2\n 넒이는 x(x-3)(x-4)(x-5) 의 양의 제곱근\n언어 : python3\n입력 설명 :\n\n다음과 같이 삼각형의 세변의 길이를 입력합니다.\n======================\n삼각형 세변의 길이(comma로 구분): 3,4,5\n출력 설명 :\n\n다음과 같이 삼각형의 넓이를 출력합니다.\n======================\n6.0\n\n샘플 입력 : 3,4,5\n샘플 출력 : 6.0\n```", "_____no_output_____" ] ], [ [ "a,b,c = eval(input())\nx = (a+b+c)/2\narea = (x*(x-a)*(x-b)*(x-c))**(0.5)\nprint(area)", "_____no_output_____" ] ], [ [ "<img src='./img/C1.jpg'> \n", "_____no_output_____" ], [ "## Example : 5\n### for 를 사용해서 암컷 개를 찾으세요. and를 사용해서 일치하는것을 찾으세요 ", "_____no_output_____" ] ], [ [ "suspects = [['낙타', '포유류','암컷'], ['상어','어류','숫컷'], ['푸들','개','암컷']]\nfor suspect in suspects:\n if suspect[1] == '개' and suspect[2] =='암컷':\n print('범인은', suspect[0], '입니다')", "_____no_output_____" ] ], [ [ "## Example : 6\n### 중복되지 않는 카드 두 장을 뽑도록 빈칸을 채우세요. ", "_____no_output_____" ] ], [ [ "import random\ncities = ['서울','부산','울산','인천' ]\nprint(random.sample(cities, 2))\n", "_____no_output_____" ] ], [ [ "## Example : 7\n### 다음중 하나를 무작위로 뽑아주세요! \nannimals = 얼룩말, 황소, 개구리, 참새.\n", "_____no_output_____" ] ], [ [ "#리스트[]\nimport random\nannimals = ['얼룩말','황소', '개구리', '참새']\nprint(random.choice(annimals))\n", "_____no_output_____" ] ], [ [ "## Example : 8\n### def 를 이용해서 서로에게 인사하는 문구를 만들어 보세요! \n가브리엘 님 안녕하세요? \\\n엘리스 님 안녕하세요? ", "_____no_output_____" ] ], [ [ "def welcome(name):\n print(name,'님 안녕하세요?')\n \nwelcome('가브리엘')\nwelcome('엘리스')\n", "_____no_output_____" ] ], [ [ "## Example : 9\n### 점수에 따라 학점을 출력 해주세요. \n철수의 점수는 75점 입니다. 몇 학점 인지 표시해 주세요. \nA학점은 80< score <=100 \nB학점은 60< score <=80 \nC학점은 40< score <=60 ", "_____no_output_____" ] ], [ [ "score =75\nif 80< score <=100:\n print('학점은 A 입니다')\nif 60< score <=80:\n print('학점은 B 입니다')\nif 40< score <=60:\n print('학점은 C 입니다')\n", "_____no_output_____" ] ], [ [ "## Example : 10\n### 변수를 사용해서 매출액을 계산해 주세요. \n주문서1 - 커피2잔, 홍차4잔, 레몬티5잔 \n주문서2 - 커피1잔, 홍차1잔, 레몬티5잔 \n주문서3 - 커피2잔, 홍차3잔, 레몬티1잔 ", "_____no_output_____" ] ], [ [ "coffee =4000\ntea = 3000\nlemon =200\norder1 = (coffee*2 + tea*4 + lemon*5)\norder2 = (coffee*1 + tea*1 + lemon*5)\norder3 = (coffee*2 + tea*3 + lemon*1)\nprint(order1+order2+order3)\n", "_____no_output_____" ] ], [ [ "## Example : 11\n### 5바퀴를 도는 레이싱 경주를 하고 있습니다. while 코드를 이용해서 트랙의 수를 카운트하고 5바퀴를 돌면 종료 멧세지를 주세요. \n반복할 때마다 몇 번째 바퀴인지 출력하세요. \\\n5바퀴를 돌면 종료 멧세지와 함께 종료해 주세요. \n", "_____no_output_____" ] ], [ [ "count = 0 \nwhile count <5:\n count =count +1\n print(count, \"번째 바퀴입니다.\")\nprint('경주가 종료되었습니다!')", "_____no_output_____" ] ], [ [ "## Example : 12\n### 정답을 맟춰보세요. \n미국이 수도는 어디인기요? \\\n보기에서 찾아서 답하게 하세요. \n 런던,오타와, 파리, 뉴욕\n틀린 답을 말하면 어느 나라의 수도인지 말해주세요. \n", "_____no_output_____" ] ], [ [ "while True:\n answer = input('런던,오타와,파리,뉴욕 중 미국이 수도는 어디일까요?')\n if answer == '뉴욕':\n print('정답입니다. 뉴욕은 미국의 수도 입니다')\n break\n elif answer == '오타와':\n print('오타와는 캐나다의 수도 입니다')\n elif answer == '파리':\n print('파리는 프랑스의 수도 입니다')\n elif answer == '런던':\n print('런던은 영국의 수도 입니다')\n else:\n print('보기에서 골라주세요')\n ", "_____no_output_____" ] ], [ [ "## Example : 13\n### 물건을 교환 해주세요 \n철수는 마트에서 형광등을 샀습니다. 그런데 LED 전구가 전기 효율이 좋아 형광등을 LED 전구로 교환 하고자 합니다. \n형광등 3개를 LED 3개로 바꾸어 주세요. \n형광등, 형광등, 형광등 ==> LED 전구, LED 전구, LED전구", "_____no_output_____" ] ], [ [ "전구 = ['형광등', '형광등', '형광등']\nfor i in range(3):\n 전구[i] = 'LED 전구'\nprint(전구)\n ", "_____no_output_____" ] ], [ [ "## Example : 14\n### 반복하기 \n동물원 원숭이 10 마리에게 인사하기. \nfor을 사용해서 10마리에게 한번에 인사하기 코드를 적어주세요.\n", "_____no_output_____" ] ], [ [ "for num in range(10):\n print ('안녕 원숭이', num) ", "_____no_output_____" ], [ "my_str ='My name is %s' % 'Lion'\nprint(my_str)", "_____no_output_____" ], [ "'%d %d' % (1,2)", "_____no_output_____" ], [ "'%f %f' % (1,2)", "_____no_output_____" ] ], [ [ "### print Options", "_____no_output_____" ] ], [ [ "print('집단지성', end='/')", "_____no_output_____" ], [ "print('집단지성', end='통합하자')", "_____no_output_____" ] ] ]

[ "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code" ]

[ [ "markdown", "markdown" ], [ "code", "code", "code" ], [ "markdown" ], [ "code", "code", "code" ], [ "markdown" ], [ "code" ], [ "markdown" ], [ "code" ], [ "markdown", "markdown" ], [ "code" ], [ "markdown" ], [ "code" ], [ "markdown" ], [ "code" ], [ "markdown" ], [ "code" ], [ "markdown" ], [ "code" ], [ "markdown" ], [ "code" ], [ "markdown" ], [ "code" ], [ "markdown" ], [ "code" ], [ "markdown" ], [ "code" ], [ "markdown" ], [ "code", "code", "code", "code" ], [ "markdown" ], [ "code", "code" ] ]

[ "MIT" ]

[ "MIT" ]

[ "MIT" ]

[ [ [ "import numpy as np\nfrom matplotlib import pylab as pl", "_____no_output_____" ], [ "import WD, WM, WMP2\nimport WP1\nimport WP2\nimport WS1\nimport WS2\nimport WS1D", "_____no_output_____" ], [ "print WD.__doc__", "This module 'WD' is auto-generated with f2py (version:2).\nFunctions:\n wd = calcwd(i,j,y,target)\n.\n" ], [ "print WD.calcwd(0.1,0.1,1.0,'Xe131')\nprint WM.calcwm(0.1,0.1,1.0,'Xe131')\nprint WMP2.calcwmp2(0.1,0.1,1.0,'Xe131')\nprint WP1.calcwp1(0.1,0.1,1.0,'Xe131')\nprint WP2.calcwp2(0.1,0.1,1.0,'Xe131')\nprint WS1.calcws1(0.1,0.1,1.0,'Xe131')\nprint WS2.calcws2(0.1,0.1,1.0,'Xe131')\nprint WS1D.calcws1d(0.1,0.1,1.0,'Xe131')", "0.00101607560646\n0.915474534035\n0.174012035131\n9.44748279608e-10\n0.0331197045743\n0.000788004428614\n0.00235111522488\n-0.000667707354296\n" ], [ "yvals = np.linspace(0, 2,100)\ny = 0.1\ntau1 = np.array([0,1])\ntau2 = np.array([0,1])\n\nX = np.vectorize(WD.calcwd)(0.1,0.1,yvals,'Xe131')\nWMthingy = np.vectorize(WM.calcwm)(tau1[0], tau2[0], y, 'Xe131')\nprint WMthingy", "736.724121094\n" ], [ "cp=1\ncn=1\nc_0 = 0.5*(cp + cn)\nprint cp", "1\n" ], [ "def calcWM(E, y, target=\"Xe131\", cp=1, cn=1):\n c = np.array([0.5*(cp + cn), 0.5*(cp - cn)])\n tau1 = np.array([0.,1.])\n tau2 = np.array([0.,1.])\n WMvals = []\n for i in range(0,2):\n for j in range(0,2):\n print tau1[i], tau2[j], c[i], c[j]\n WMvals.append(c[i]*c[j]*WM.calcwm(tau1[i], tau2[j], y, target))\n return sum(WMvals)\n\nprint calcWM(0.1, 0.1, \"Xe131\")", "0.0 0.0 1.0 1.0\n0.0 1.0 1.0 0.0\n1.0 0.0 0.0 1.0\n1.0 1.0 0.0 0.0\n736.724121094\n" ], [ "yvals = np.linspace(0, 2,100)\n\npl.figure()\n# pl.plot(yvals, np.vectorize(WD.calcwd)(0.1,0.1,yvals,'Xe131'))\n# pl.plot(yvals, np.vectorize(WM.calcwm)(0.1,0.1,yvals,'Xe131'))\n# pl.plot(yvals, np.vectorize(WMP2.calcwmp2)(0.1,0.1,yvals,'Xe131'))\n# pl.plot(yvals, np.vectorize(WP1.calcwp1)(0.1,0.1,yvals,'Xe131'))\n# pl.plot(yvals, np.vectorize(WP2.calcwp2)(0.1,0.1,yvals,'Iodine'))\n# pl.plot(yvals, np.vectorize(WS1.calcws1)(0.1,0.1,yvals,'Xe131'))\n# pl.plot(yvals, np.vectorize(WS2.calcws2)(0.1,0.1,yvals,'Xe131'))\n# pl.plot(yvals, np.vectorize(WS1D.calcws1d)(0.1,0.1,yvals,'Flourine'))\npl.show()", "_____no_output_____" ] ] ]

[ "code" ]

[ [ "code", "code", "code", "code", "code", "code", "code", "code" ] ]

[ "CNRI-Python" ]

[ "CNRI-Python" ]

[ "CNRI-Python" ]

[ [ [ "!pip install requests\n!pip install bs4\nimport requests\nimport bs4\n\nfrom google.colab import drive\n\ndrive.mount('/gdrive')\n\nimport pandas\nfrom bs4 import BeautifulSoup\nimport requests\nimport time", "Requirement already satisfied: requests in /usr/local/lib/python3.7/dist-packages (2.23.0)\nRequirement already satisfied: idna<3,>=2.5 in /usr/local/lib/python3.7/dist-packages (from requests) (2.10)\nRequirement already satisfied: urllib3!=1.25.0,!=1.25.1,<1.26,>=1.21.1 in /usr/local/lib/python3.7/dist-packages (from requests) (1.24.3)\nRequirement already satisfied: chardet<4,>=3.0.2 in /usr/local/lib/python3.7/dist-packages (from requests) (3.0.4)\nRequirement already satisfied: certifi>=2017.4.17 in /usr/local/lib/python3.7/dist-packages (from requests) (2021.10.8)\nRequirement already satisfied: bs4 in /usr/local/lib/python3.7/dist-packages (0.0.1)\nRequirement already satisfied: beautifulsoup4 in /usr/local/lib/python3.7/dist-packages (from bs4) (4.6.3)\n" ], [ "# UBER PART\nbase_url = 'https://www.consumeraffairs.com/travel/uber.html?page={}'\n\ncomment_list = []\ncomment_list = set(comment_list)\ncounter = 0\n\nwhile True:\n counter += 1\n\n res = requests.get(base_url.format(counter))\n soup = BeautifulSoup(res.text, 'html.parser')\n\n for item in soup.select('div[class=\"rvw-bd\"] p p'):\n if item.getText() not in comment_list:\n comment_list.add(item.getText())\n\n if counter == 30:\n break\n time.sleep(3)\n\n\nprint(len(comment_list))\n\n\ndf = pandas.DataFrame(comment_list)\nwith open('/gdrive/My Drive/uber_data.csv', 'w') as f:\n df.to_csv(f)", "Requirement already satisfied: requests in /usr/local/lib/python3.6/dist-packages (2.23.0)\nRequirement already satisfied: idna<3,>=2.5 in /usr/local/lib/python3.6/dist-packages (from requests) (2.10)\nRequirement already satisfied: certifi>=2017.4.17 in /usr/local/lib/python3.6/dist-packages (from requests) (2020.11.8)\nRequirement already satisfied: urllib3!=1.25.0,!=1.25.1,<1.26,>=1.21.1 in /usr/local/lib/python3.6/dist-packages (from requests) (1.24.3)\nRequirement already satisfied: chardet<4,>=3.0.2 in /usr/local/lib/python3.6/dist-packages (from requests) (3.0.4)\nRequirement already satisfied: bs4 in /usr/local/lib/python3.6/dist-packages (0.0.1)\nRequirement already satisfied: beautifulsoup4 in /usr/local/lib/python3.6/dist-packages (from bs4) (4.6.3)\nDrive already mounted at /gdrive; to attempt to forcibly remount, call drive.mount(\"/gdrive\", force_remount=True).\n842\n" ], [ "# TAXI PART\nbase_url = 'https://www.reviews.co.uk/company-reviews/store/airports-taxi-transfers/{}'\n\ntaxi_comment_list = []\ntaxi_comment_list = set(taxi_comment_list)\ncounter = 0\n\nwhile True:\n counter += 1\n\n res = requests.get(base_url.format(counter))\n soup = BeautifulSoup(res.text, 'html.parser')\n\n for item in soup.select('span[class=\"Review__body\"]'):\n if item.getText() not in taxi_comment_list:\n taxi_comment_list.add(item.getText()[1:-1])\n\n if counter == 78:\n break\n time.sleep(3)\n\n\nprint(len(taxi_comment_list))\n\n\ndf_taxi = pandas.DataFrame(taxi_comment_list)\nwith open('/gdrive/My Drive/taxi_data.csv', 'w') as f:\n df_taxi.to_csv(f)", "1296\n" ], [ "# Concatenate both .csv files into one and save it to new .csv\nvertical_stack = pandas.concat([df_taxi, df], axis=0)\n\nwith open('/gdrive/My Drive/vertical_stack.csv', 'w') as f:\n vertical_stack.to_csv(f)", "_____no_output_____" ], [ "", "_____no_output_____" ] ] ]

[ "code" ]

[ [ "code", "code", "code", "code", "code" ] ]

[ "IJG" ]

[ "IJG" ]

[ "IJG" ]

[ [ [ "#Numerical Differentiation\n#part 2a)\n\nimport numpy as np\nimport matplotlib.pyplot as plt\nfrom scipy.interpolate import interp1d, barycentric_interpolate\n\ndef diff(f,x,h=0.00000000001):\n return (f(x+h)-f(x-h))/2/h\n\nx = [0. , 0.1 , 0.2 , 0.3 , 0.4 ]\ny = [0.000000,0.078348,0.138910,0.192916,0.244981]\n\nf_cubic = interp1d(x,y, kind = 'cubic')\n\nx0 = np.arange(0.,0.4,0.001)\ny0 = barycentric_interpolate(x,y,x0)\nz = np.arange(0.1,0.3,0.001)\nz0 = diff(f_cubic,z)\nplt.plot(x,y,'kx')\nplt.plot(x0,f_cubic(x0), 'r-')\nplt.show()\n\nplt.plot(z,z0,'k-', markersize = 10)\nplt.plot(0.2,diff(f_cubic,0.2),'ro',markersize=6, label = \"f'(0.2)\")\nplt.legend()\nplt.show()\nprint (diff(f_cubic,0.2))", "_____no_output_____" ] ], [ [ "$part \\ 2b)$", "_____no_output_____" ] ], [ [ "#Define Subroutines\ndef f(x):\n return np.sin(x) + np.exp(-x)\ndef df(x):\n return -np.sin(x) + np.exp(-x)\ndef g(x,h):\n return 1/h**2*(f(x+h)-2*f(x)+f(x-h))\ndef G(x,h, p):\n G = (2**p*g(x,h/2) - g(x,h))/(2**p-1)\n return G\n#\n\n#Computation\nt = np.arange(0,10,0.1)\nfor h in [0.1, 0.5]:\n plt.plot(t,f(t) ,'-', color = 'black', linewidth = 3, label = \"f(x)\")\n plt.plot(t,df(t) ,'-', color = 'grey' , linewidth = 2, label = \"f'(x)\")\n plt.plot(t,g(t,h) ,'--', color = 'red' , linewidth = 2, label = \"g(x)\")\n plt.plot(t,G(t,h, 2),'-.', color = 'green', linewidth = 2, label = \"G(x)\")\n plt.title(\"h = \" + str(h))\n plt.ylabel(\"y axis\")\n plt.xlabel(\"x axis\")\n plt.legend()\n plt.show()\n#\n\n#Output\n#", "_____no_output_____" ] ] ]

[ "code", "markdown", "code" ]

[ [ "code" ], [ "markdown" ], [ "code" ] ]

[ "BSD-3-Clause" ]

[ "BSD-3-Clause" ]

[ "BSD-3-Clause" ]

[ [ [ "## Embedding a Bokeh server in a Notebook\n\nThis notebook shows how a Bokeh server application can be embedded inside a Jupyter notebook. ", "_____no_output_____" ] ], [ [ "import yaml\n\nfrom bokeh.layouts import column\nfrom bokeh.models import ColumnDataSource, Slider\nfrom bokeh.plotting import figure\nfrom bokeh.themes import Theme\nfrom bokeh.io import show, output_notebook\n\nfrom bokeh.sampledata.sea_surface_temperature import sea_surface_temperature\n\noutput_notebook()", "_____no_output_____" ] ], [ [ "There are various application handlers that can be used to build up Bokeh documents. For example, there is a `ScriptHandler` that uses the code from a `.py` file to produce Bokeh documents. This is the handler that is used when we run `bokeh serve app.py`. In the notebook we can use a function to define a Bokehg application.\n\nHere is the function `bkapp(doc)` that defines our app:", "_____no_output_____" ] ], [ [ "def bkapp(doc):\n df = sea_surface_temperature.copy()\n source = ColumnDataSource(data=df)\n\n plot = figure(x_axis_type='datetime', y_range=(0, 25),\n y_axis_label='Temperature (Celsius)',\n title=\"Sea Surface Temperature at 43.18, -70.43\")\n plot.line('time', 'temperature', source=source)\n\n def callback(attr, old, new):\n if new == 0:\n data = df\n else:\n data = df.rolling('{0}D'.format(new)).mean()\n source.data = ColumnDataSource.from_df(data)\n\n slider = Slider(start=0, end=30, value=0, step=1, title=\"Smoothing by N Days\")\n slider.on_change('value', callback)\n\n doc.add_root(column(slider, plot))\n\n doc.theme = Theme(json=yaml.load(\"\"\"\n attrs:\n Figure:\n background_fill_color: \"#DDDDDD\"\n outline_line_color: white\n toolbar_location: above\n height: 500\n width: 800\n Grid:\n grid_line_dash: [6, 4]\n grid_line_color: white\n \"\"\", Loader=yaml.FullLoader))", "_____no_output_____" ] ], [ [ "Now we can display our application using ``show``, which will automatically create an ``Application`` that wraps ``bkapp`` using ``FunctionHandler``. The end result is that the Bokeh server will call ``bkapp`` to build new documents for every new sessions that is opened.\n\n**Note**: If the current notebook is not displayed at the default URL, you must update the `notebook_url` parameter in the comment below to match, and pass it to `show`.", "_____no_output_____" ] ], [ [ "show(bkapp) # notebook_url=\"http://localhost:8888\" ", "_____no_output_____" ] ] ]

[ "markdown", "code", "markdown", "code", "markdown", "code" ]

[ [ "markdown" ], [ "code" ], [ "markdown" ], [ "code" ], [ "markdown" ], [ "code" ] ]

[ "BSD-3-Clause" ]

[ "BSD-3-Clause" ]

[ "BSD-3-Clause" ]

[ [ [ "import numpy as np\nfrom sklearn.neighbors import NearestNeighbors", "_____no_output_____" ], [ "r = 1 # Initial radius for model\nx = 1 # increment for expanding radius\nd = .01 # distance threshold for determing if two points near equidistant\n\n# existing points and corresponding classes\nexistingPoints = np.array([[2,3,4,5,6,7,8],[1,3,4,5,6,7,8],[6,2,3,4,5,6,7]]) # existing points in model\nexistingClasses = np.array([1,2,3])\n\n# initialize classifier\nneigh = NearestNeighbors(radius = r)\n\n# read in new point\nnewPoints = np.array([2,3,4,5,6,7,8],ndmin=2)\n\n# fit existing points to classifier\nneigh.fit(existingPoints)\n\n#distances,indices = neigh.radius_neighbors(newPoints)\n\nnewClass = classify(neigh,existingClasses,newPoints)\n\n# Add the new point and new class to the model. Class = 0 if point is unclassified\nexistingPoints = np.append(existingPoints,newPoints)\nexistingClasses = np.append(existingClasses,newClass)\n\nprint(newClass)", "2\n" ], [ "def classify(neigh,existingClasses,newPoints):\n \n distances,indices = neigh.radius_neighbors(newPoints)\n\n # if there are no points in the radius, expand radius by x and check again before classifying point.\n if len(indices)==0 or len(indices)==1: \n neigh.radius = r+x\n distances,indices = neigh.radius_neighbors(newPoints) \n \n # else if there are two points from different classes that are close to the same distance\n # (within distance threshold), expand radius to see if there is another very close point\n elif len(indices) ==2 and existingClasses[indices[0]]!=existingClasses[indices[1]]:\n if abs(distances[0]-distances[1]) <= d:\n neigh.radius = r+x\n \n # predict class of new data \n if len(indices)!=0:\n # calculate weights (arbitrary weights for now)\n weights = np.array([0,5,5])\n \n # sum weights for each class\n classes = existingClasses[indices[0]]\n classes = np.unique(classes[np.where(classes!=0)]) # ignore zero class (outliers)\n \n classWeight = np.zeros(len(classes)) # initialize weight array\n \n for i,cl in enumerate(classes):\n classWeight[i] = sum(weights[np.where(classes==cl)])\n \n newClass = classes[np.argmax(classWeight)] \n \n else: \n newClass = 0\n \n return(newClass)\n \n \n\n\n\n\n", "_____no_output_____" ], [ "## Since we are including PAMguard into feature space, use an enum to make np.array work", "_____no_output_____" ], [ "from datetime import datetime\n\nclass DetectedTarget:\n \n features_label = [\"\",\"\",\"\",\"\",\"\",\"\",\"\",\"\"]\n \n def __init__(self, features=[], source=\"\", date=datetime.now(), classification=None):\n self.features = features\n self.source = source\n self.date = date\n self.classification = classification", "_____no_output_____" ], [ "def rescaleFrom0To1(features):\n for i in range(features.shape[1]):\n if features[:,i].min() == features[:,i].max():\n features[:,i] = 0.5\n else:\n features[:,i] = (features[:,i] - features[:,i].min()) / (features[:,i].max() - features[:,i].min())\n return features", "_____no_output_____" ], [ "features", "_____no_output_____" ], [ "rescaleFrom0To1(features)", "[ 0. 1.]\n[ 0. 1.]\n[ 0. 1.]\n[ 0.5 0.5]\n[ 1. 0.]\n[ 1. 0.]\n[ 1. 0.]\n" ], [ "from enum import Enum\nimport numpy as np\nfrom sklearn.neighbors import NearestNeighbors\nfrom datetime import datetime\nimport os.path\nimport csv\n\nclass Classes(Enum):\n Interesting = 1.1\n NotInteresting = 1.2\n \n FastLarge = 2.1\n FastSmall = 2.2\n SlowLarge = 2.3\n SlowSmall = 2.4\n \n # SchoolOfFish = 3.1\n # SingleFish = 3.2\n # Kelp = 3.3\n # DolphinOrPorpoise = 3.4\n\nclass DetectedTarget:\n \n features_label = [\"\",\"\",\"\",\"\",\"\",\"\",\"\",\"\"]\n \n def __init__(self, features=[], source=\"Unknown\", date=datetime.now(), classification=None):\n self.features = features\n self.source = source\n self.date = date\n self.classification = classification\n\nclass weightedNeighbors:\n \n def __init__(self):\n \n # Global variables\n self.INITIAL_RADIUS = 3.0\n self.SIZE_FEATURE_WEIGHT = 1.0\n self.SPEED_FEATURE_WEIGHT = 1.0\n self.SPEED_RELATIVE_TO_CURRENT_FEATURE_WEIGHT = 1.0\n self.TARGET_STRENGTH_FEATURE_WEIGHT = 1.0\n self.CURRENT_SPEED_FEATURE_WEIGHT = 1.0\n self.TIME_OF_DAY_FEATURE_WEIGHT = 1.0\n self.PASSIVE_ACOUSTICS_FEATURE_WEIGHT = 1.0\n self.RADIUS_INCREMENT = 0.1 # increment for expanding radius\n self.DISTANCE_THRESHOLD = 0.01 # distance threshold for determing if two points near equidistant\n \n # load current model and initialize NearestNeighbors model\n self.current_model_targets = self.load_detectedTargets()\n self.model = NearestNeighbors(radius=self.INITIAL_RADIUS)\n \n def load_detectedTargets(self):\n '''\n Load existing targets from current_model_targets.csv\n '''\n \n current_model_targets = []\n \n if os.path.isfile('current_model_targets.csv'):\n current_model_targets = []\n with open('current_model_targets.csv', 'r') as f:\n reader = csv.reader(f,delimiter = \";\")\n next(reader,None)\n for target in reader:\n current_model_targets.append( DetectedTarget(\n features=list(target[1:8]), source=target[8], date=target[10],\n classification=Classes(float(target[-1])))) \n\n else:\n print('No existing targets for model')\n \n return current_model_targets\n \n def fitModel(self):\n '''\n Fit current model targets to model\n '''\n self.model.fit(np.array(list(map(lambda x: x.features, self.current_model_targets))))\n \n def determine_weights(self,indices):\n '''\n This function will return the weights of points with the desired indices\n '''\n pass\n return np.array([0,5,5])\n \n \n def classify(self,newPoint):\n '''\n Predict class of new target detection(s)\n '''\n x = self.RADIUS_INCREMENT \n r = self.DISTANCE_THRESHOLD \n \n \n existingClasses = np.array(list(map(lambda x: x.classification, self.current_model_targets)))\n\n distances,indices = self.model.radius_neighbors(newPoint)\n\n # if there are no points in the radius, expand radius by x and check again before classifying point.\n if len(indices)==0 or len(indices)==1: \n neigh.radius = r+x\n distances,indices = self.model.radius_neighbors(newPoint) \n\n # else if there are two points from different classes that are close to the same distance\n # (within distance threshold), expand radius to see if there is another very close point\n elif len(indices) ==2 and existingClasses[indices[0]]!=existingClasses[indices[1]]:\n if abs(distances[0]-distances[1]) <= d:\n neigh.radius = r+x\n distances,indices = self.model.radius_neighbors(newPoint) \n\n # predict class of new data \n if len(indices)!=0:\n # calculate weights (arbitrary weights for now)\n weights = self.determine_weights(indices)\n\n # sum weights for each class\n classes = existingClasses[indices[0]]\n classes = np.unique(classes[np.where(classes!=0)]) # ignore zero class (outliers)\n\n classWeight = np.zeros(len(classes)) # initialize weight array\n\n for i,cl in enumerate(classes):\n classWeight[i] = sum(weights[np.where(classes==cl)])\n\n newClass = classes[np.argmax(classWeight)] \n\n else: \n newClass = 0\n \n return(newClass) ", "_____no_output_____" ], [ "neigh = weightedNeighbors()\nneigh.fitModel()\nneigh.classify(np.array([2,3,4,5,6,7,8],ndmin=2))", "['7', '6', '5', '4', '3', '2', '1']\n[['1' '2' '3' '4' '5' '6' '7']\n ['7' '6' '5' '4' '3' '2' '1']]\n" ] ] ]

[ "code" ]

[ [ "code", "code", "code", "code", "code", "code", "code", "code", "code", "code" ] ]

[ "MIT" ]

[ "MIT" ]

[ "MIT" ]

[ [ [ "empty" ] ] ]

[ "empty" ]

[ [ "empty" ] ]

[ "Apache-2.0" ]

[ "Apache-2.0" ]

[ "Apache-2.0" ]

[ [ [ "# imports\n%matplotlib inline\nimport matplotlib.pyplot as plt\nimport numpy as np\n\nfrom __future__ import division", "_____no_output_____" ], [ "M2_amp_obs_SoG = 0.902", "_____no_output_____" ], [ "pha_diff = {'ene': 86,\n 'wHoll': 87,\n 'llap': 87,\n 'wslip': 87,\n 'TS4': 86,\n 'smooth': 58,\n 'downto2': 82,\n 'downto5': 83,\n 'weaklog': 89,\n 'd10': 85,\n 'u20': 86,\n 'topog1': 80,\n 'base': 91,\n 'stronglog': 101,\n 'secondlog': 96,\n 'sticky': 90,\n 'deep' : 55,\n 'deep_fric': 83,\n 'GmO': 75,\n 'GmO_TS5': 81,\n 'GmO_TS6': 81,\n 'GmO_TS7': 77,\n 'GmO_TS8': 74,\n 'GmO_TS9': 75,\n 'GmO_TS10': 76,\n 'a1': 77,\n 'a2': 76,\n 'M17_a1': 94,\n 'M17_a2': 92,\n 'M17_a3': 88,\n 'M17_a4': 81,\n 'M17_a5': 77,\n 'M17_a6': 75,\n 'M17_a7': 73,\n 'M17_a9': 68,\n 'M17_a10': 73,\n 'M17_a11': 81,\n 'M17_a12': 75,\n 'M17_a13': 79,\n 'M17_a14': 77}\namp = {'ene': 0.25,\n 'wHoll': 0.255,\n 'llap': 0.29,\n 'wslip': 0.27,\n 'TS4': 0.26,\n 'smooth': 0.18,\n 'downto2': 0.24,\n 'downto5': 0.24,\n 'weaklog': 0.3,\n 'd10': 0.255,\n 'u20': 0.26,\n 'topog1': 0.28,\n 'base': 0.33,\n 'stronglog': 0.49,\n 'secondlog': 0.39,\n 'sticky': 0.40,\n 'deep': 0.19,\n 'deep_fric': 0.34,\n 'GmO': 0.36,\n 'GmO_TS5': 0.40,\n 'GmO_TS6': 0.40,\n 'GmO_TS7': 0.37,\n 'GmO_TS8': 0.34,\n 'GmO_TS9': 0.35,\n 'GmO_TS10': 0.36,\n 'a1': 0.37,\n 'a2': 0.36,\n 'M17_a1': 0.47,\n 'M17_a2': 0.45,\n 'M17_a3': 0.41,\n 'M17_a4': 0.36,\n 'M17_a5': 0.34,\n 'M17_a6': 0.31,\n 'M17_a7': 0.31,\n 'M17_a9': 0.30,\n 'M17_a10': 0.32,\n 'M17_a11': 0.37,\n 'M17_a12': 0.33,\n 'M17_a13': 0.36,\n 'M17_a14': 0.35}", "_____no_output_____" ], [ "bathymetry2 = ('llap', 'downto5', 'downto2', 'stronglog', 'weaklog', 'wslip', 'secondlog', 'base')", "_____no_output_____" ], [ "count = 0\nfig, ax = plt.subplots(1,1,figsize=(18, 4.5))\nfor key in pha_diff:\n count += 1\n if count <= 7:\n symbol = 'o'\n elif count <=14:\n symbol = '^'\n elif count <=21:\n symbol = '>'\n else:\n symbol = '<'\n ax.plot(pha_diff[key], amp[key], symbol, label=key)\nax.plot(75,0.37,'gs', label='Obs', markersize=10)\nax.plot(pha_diff['a2'], amp['a2'], 'mo', markersize=8, label='Proper Bottom')\nax.plot(92,0.33,'*',label='corr15')\nax.set_xlim((33,105))\nax.legend(loc='upper left')\nax.set_ylabel('Amplitude at Victoria (m)')\nax.set_xlabel('Phase difference Victoria to Pt Atkinson (deg)')", "_____no_output_____" ], [ "print (pha_diff['GmO'] + pha_diff['downto2'] - pha_diff['TS4'], 75, pha_diff['GmO_TS6']\n )\nprint (amp['GmO'] + amp['downto2'] - amp['TS4'], 0.37, amp['GmO_TS6']\n )\n", "71 75 81\n0.33999999999999997 0.37 0.4\n" ], [ "fig, ax = plt.subplots(1,1,figsize=(9,9))\nfor key in bathymetry2:\n ax.plot(pha_diff[key], amp[key]/M2_amp_obs_SoG, symbol, label=key, markersize=10)\nax.plot(75,0.37/M2_amp_obs_SoG,'s',label='Obs', markersize=20)\nax.plot(92,0.33/M2_amp_obs_SoG,'o',label='Nowcast v1.0', markersize=15)\nax.set_xlim((55,105))\nax.tick_params(axis='x', labelsize=16)\nax.tick_params(axis='y', labelsize=16)\nax.legend(loc='upper left', fontsize=16)\nax.set_ylabel('Amplitude Ratio at Victoria to Pt Atkinson', fontsize=16)\nax.set_xlabel('Phase difference Victoria to Pt Atkinson (deg)', fontsize=16)", "_____no_output_____" ], [ "fig, ax = plt.subplots(1,1,figsize=(9,9))\nfor key in bathymetry2:\n ax.plot(pha_diff[key], amp[key]/M2_amp_obs_SoG, '>b', markersize=10)\nfor key in ('smooth', 'sticky'):\n ax.plot(pha_diff[key], amp[key]/M2_amp_obs_SoG, '>m', markersize=10)\nfor key in ('deep', 'deep_fric' ):\n ax.plot(pha_diff[key], amp[key]/M2_amp_obs_SoG, '>y', markersize=10)\nax.plot(pha_diff['GmO'], amp[\"GmO\"]/M2_amp_obs_SoG, 'ro', markersize=15, label = 'New Bathymetry')\nax.plot(pha_diff['a1'], amp['a1']/M2_amp_obs_SoG, 'm*', markersize=10, label='Proper Bottom')\nax.plot(pha_diff['a2'], amp['a2']/M2_amp_obs_SoG, 'mo', markersize=10, label='Proper Bottom 2')\n\nax.plot(75,0.37/M2_amp_obs_SoG,'sg',label='Obs', markersize=20)\n#ax.plot(92,0.33/M2_amp_obs_SoG,'or',label='Nowcast v1.0', markersize=15)\nax.set_xlim((53,105))\nax.tick_params(axis='x', labelsize=16)\nax.tick_params(axis='y', labelsize=16)\nax.legend(loc='upper left', fontsize=16)\nax.set_ylabel('Amplitude Ratio at Victoria to Pt Atkinson', fontsize=16)\nax.set_xlabel('Phase difference Victoria to Pt Atkinson (deg)', fontsize=16)", "_____no_output_____" ], [ "fig, ax = plt.subplots(1,1,figsize=(9,9))\nfor key in ['GmO', 'GmO_TS5', 'GmO_TS6', 'GmO_TS7', 'GmO_TS8', 'GmO_TS9', 'GmO_TS10']:\n ax.plot(pha_diff[key], amp[key]/M2_amp_obs_SoG, '>', markersize=10, label=key)\nax.plot(75,0.37/M2_amp_obs_SoG,'sg',label='Obs', markersize=20)\nax.plot(pha_diff['a1'], amp['a1']/M2_amp_obs_SoG, 'm*', markersize=10, label='Proper Bottom')\nax.plot(pha_diff['a2'], amp['a2']/M2_amp_obs_SoG, 'mo', markersize=10, label='Proper Bottom 2')\n\nax.set_xlim((74,82))\nax.tick_params(axis='x', labelsize=16)\nax.tick_params(axis='y', labelsize=16)\nax.legend(loc='lower right', fontsize=16)\nax.set_ylabel('Amplitude Ratio at Victoria to Pt Atkinson', fontsize=16)\nax.set_xlabel('Phase difference Victoria to Pt Atkinson (deg)', fontsize=16)", "_____no_output_____" ], [ "6/40.*52\n", "_____no_output_____" ] ] ]

[ "code" ]

[ [ "code", "code", "code", "code", "code", "code", "code", "code", "code", "code" ] ]

[ "MIT" ]

[ "MIT" ]

[ "MIT" ]

[ [ [ "# What is Abstraction in OOP", "_____no_output_____" ], [ "<ul><li>Abstraction is the concept of object-oriented programming that “shows” only essential attributes and “hides” unnecessary information.</li><li>The main purpose of abstraction is hiding the unnecessary details from the users. </li><li> Abstraction is selecting data from a larger pool to show only relevant details of the object to the user. </li><li> It helps in reducing programming complexity and efforts. </li><li>It is one of the most important concepts of OOPs.</li></ul> ", "_____no_output_____" ], [ "# Abstraction in Python", "_____no_output_____" ], [ "<ul><li>Abstraction in python is defined as hiding the implementation of logic from the client and using the particular application. </li><li>It hides the irrelevant data specified in the project, reducing complexity and giving value to the efficiency.</li><li> Abstraction is made in Python using <b>Abstract classes</b> and their methods in the code.</li></ul>", "_____no_output_____" ], [ "## What is an Abstract Class?", "_____no_output_____" ], [ "<ul><li>Abstract Class is a type of class in OOPs, that declare one or more abstract methods. </li><li>These classes can have abstract methods as well as concrete methods. </li><li>A normal class cannot have abstract methods.</li><li>An abstract class is a class that contains at least one abstract method.</li></ul>", "_____no_output_____" ], [ "## What are Abstract Methods?\n<ul><li>Abstract Method is a method that has just the method definition but does not contain implementation.</li><li>A method without a body is known as an Abstract Method.</li><li>It must be declared in an abstract class.</li><li>The abstract method will never be final because the abstract class must implement all the abstract methods.</li></ul>", "_____no_output_____" ], [ "## When to use Abstract Methods & Abstract Class?\n<ul><li>Abstract methods are mostly declared where two or more subclasses are also doing the same thing in different ways through different implementations.</li><li>It also extends the same Abstract class and offers different implementations of the abstract methods.</li><li>Abstract classes help to describe generic types of behaviors and object-oriented programming class hierarchy. </li><li>It also describes subclasses to offer implementation details of the abstract class.</li></ul>", "_____no_output_____" ], [ "## Difference between Abstraction and Encapsulation", "_____no_output_____" ], [ "<table style=\"background-color:#ffe6e6\">\n <tr><th><b>Abstraction</b></th><th><b>Encapsulation</b></th></tr>\n <tr><td>Abstraction in Object Oriented Programming solves the issues at the design level.</td><td>Encapsulation solves it implementation level.</td></tr>\n <tr><td>Abstraction in Programming is about hiding unwanted details while showing most essential information.</td><td>Encapsulation means binding the code and data into a single unit.</td></tr>\n <tr><td>Data Abstraction in Java allows focussing on what the information object must contain</td><td>Encapsulation means hiding the internal details or mechanics of how an object does something for security reasons.</td></tr>\n</table>", "_____no_output_____" ], [ "## Advantages of Abstraction\n<ol><li>The main benefit of using an Abstraction in Programming is that it allows you to group several related classes as siblings.</li><li>\nAbstraction in Object Oriented Programming helps to reduce the complexity of the design and implementation process of software.</li></ol>", "_____no_output_____" ], [ "## How Abstract Base classes work : \n<ul><li>By default, Python does not provide abstract classes. Python comes with a module that provides the base for defining Abstract Base classes(ABC) and that module name is ABC. </li><li>ABC works by decorating methods of the base class as abstract and then registering concrete classes as implementations of the abstract base. </li><li>A method becomes abstract when decorated with the keyword @abstractmethod.</li></ul>", "_____no_output_____" ], [ "#### Syntax\n\nAbstract class Syntax is declared as:", "_____no_output_____" ] ], [ [ "from abc import ABC\n\n# declaration\nclass classname(ABC):\n def pau(self):\n pass\n \n ", "_____no_output_____" ] ], [ [ "Abstract method Syntax is declared as", "_____no_output_____" ] ], [ [ "def abstractmethod_name():\n pass\n ", "_____no_output_____" ] ], [ [ "### Few things to be noted in Python:\n\n<ul><li>In python, an abstract class can hold both an abstract method and a normal method.</li><li>\nThe second point is an abstract class is not initiated (no objects are created).</li><li>\nThe derived class implementation methods are defined in abstract base classes.</li></ul>", "_____no_output_____" ] ], [ [ "from ABC import abc\n\n# here abc and ABC are case-sensitive. When we swap it creates", "_____no_output_____" ] ], [ [ "### Code I:", "_____no_output_____" ] ], [ [ "from abc import ABC, abstractmethod\n\n# Abstract Class\nclass product(abc): \n \n # Normal Method\n def item_list(self, rate):\n print(\"amount submitted : \",rate)\n \n # Abstract Method\n @abstractmethod\n def product(self, rate): \n ", "_____no_output_____" ] ], [ [ "### Code II:\nA program to generate the volume of geometric shapes", "_____no_output_____" ] ], [ [ "from abc import ABC\n\nclass geometric(ABC):\n \n def volume(self):\n #abstract method\n pass\n \nclass Rect(geometric):\n length = 4\n width = 6\n height = 6\n \n def volume(self):\n return self.length * self.width *self.height\n \nclass Sphere(geometric):\n radius = 8\n def volume(self):\n return 1.3 * 3.14 * self.radius * self.radius *self.radius\n \nclass Cube(geometric):\n Edge = 5\n def volume(self):\n return self.Edge * self.Edge *self.Edge\n \nclass Triangle_3D:\n length = 5\n width = 4\n def volume(self):\n return 0.5 * self.length * self.width\n \nrr = Rect()\nss = Sphere()\ncc = Cube()\ntt = Triangle_3D()\nprint(\"Volume of a rectangle:\", rr.volume())\nprint(\"Volume of a circle:\", ss.volume())\nprint(\"Volume of a square:\", cc.volume())\nprint(\"Volume of a triangle:\", tt.volume())", "Volume of a rectangle: 144\nVolume of a circle: 2089.9840000000004\nVolume of a square: 125\nVolume of a triangle: 10.0\n" ] ], [ [ "### Code III\nA program to generate different invoices", "_____no_output_____" ] ], [ [ "from abc import ABC, abstractmethod\n\nclass Bill(ABC):\n def final_bill(self, pay):\n print('Purchase of the product: ', pay)\n \n @abstractmethod\n def Invoice(self, pay):\n pass\n \nclass Paycheque(Bill):\n def Invoice(self, pay):\n print('paycheque of: ', pay)\n \nclass CardPayment(Bill):\n def Invoice(self, pay):\n print('pay through card of: ', pay)\n \naa = Paycheque()\naa.Invoice(6500)\naa.final_bill(6500)\nprint(isinstance(aa,Invoice))\naa = CardPayment()\naa.Invoice(2600)\naa.final_bill(2600)\nprint(isinstance(aa,Invoice))", "_____no_output_____" ] ], [ [ "### Code IV:\n Python program showing abstract base class work", "_____no_output_____" ] ], [ [ "from abc import ABC, abstractmethod\n\nclass Animal(ABC):\n\n @abstractmethod\n def move(self):\n pass\n\nclass Human(Animal):\n \n def move(self):\n print(\"I can walk and run\")\n\nclass Snake(Animal):\n \n def move(self):\n print(\"I can crawl\")\n\nclass Dog(Animal):\n\n def move(self):\n print(\"I can bark\")\n\nclass Lion(Animal):\n \n def move(self):\n print(\"I can roar\")\n\n# Object Instantiation\nr = Human()\nr.move()\n\nk = Snake()\nk.move()\n\nd = Dog()\nd.move()\n\nm = Lion()\nm.move()\n", "I can walk and run\nI can crawl\nI can bark\nI can roar\n" ] ], [ [ "### Concrete Methods in Abstract Base Classes : \n<ul><li>Concrete (normal) classes contain only concrete (normal) methods whereas abstract classes may contain both concrete methods and abstract methods.</li><li> The concrete class provides an implementation of abstract methods, the abstract base class can also provide an implementation by invoking the methods via super().</li></ul>", "_____no_output_____" ], [ "### Code V:\nPython program invoking a method using super()", "_____no_output_____" ] ], [ [ "from abc import ABC, abstractmethod\n\nclass R(ABC):\n \n def rk(self):\n print(\"Abstract Base Class\")\n\nclass K(R):\n def rk(self):\n super().rk()\n print(\"subclass\")\n\n# Object instantiation\nr = K()\nr.rk()\n", "Abstract Base Class\nsubclass\n" ] ], [ [ "### Code VI:", "_____no_output_____" ] ], [ [ "from abc import ABC, abstractmethod\n\nclass Bank(ABC):\n def branch(self, Naira):\n print(\"Fees submitted : \",Naira)\n \n @abstractmethod\n def Bank(Naira):\n \n \nclass private(Bank):\n def Bank(naira):\n print(\"Total Naira Value here: \",Naira)\n \nclass public(bank):\n def Bank(Naira):\n print(\"Total Naira Value here:\",Naira)\n\nprivate.Bank(5000)\npublic.Bank(2000)\n\na = public()\n#a.branch(3500)", "_____no_output_____" ] ], [ [ "## Class Project I", "_____no_output_____" ], [ "Develop a python OOP program that creates an abstract base class called coup_de_ecriva. The base class will have one abstract method called <b>Fan_Page</b> and four subclassses namely; <b>FC_Cirok, Madiba_FC, Blue_Jay_FC and TSG_Walker</b>. The program will receive as input the name of the club the user supports and instantiate an object that will invoke the <b>Fan_Page</b> method in the subclass that prints Welcome to <b>\"club name\"</b>.\n\n<p><b>Hint:</b></p>\nThe subclasses will use <b>Single Inheritance</b> to inherit the abstract base class.\n ", "_____no_output_____" ] ], [ [ "from abc import ABC, abstractmethod\n\nclass coup_de_escriva(ABC):\n \n @abstractmethod\n def Fan_page(self):\n pass\n \nclass FC_Cirok(coup_de_escriva):\n def Fan_page(self):\n print(str(input(\"Enter your name\")))\n print(str(input(\"Which club do you support?\")))\n print(\"WELCOME TO CIROK FC!\")\n\nclass Madiba_FC(coup_de_escriva):\n def Fan_page(self):\n print(str(input(\"Enter your name\")))\n print(str(input(\"Which club do you support?\")))\n print(\"WELCOME TO MADIBA FC!\")\n\n \nclass Blue_Jay_FC(coup_de_escriva):\n def Fan_page(self):\n print(str(input(\"Enter your name\")))\n print(str(input(\"Which club do you support?\")))\n print(\"WELCOME TO THE BLUES!\")\n \nclass TSG_Walkers(coup_de_escriva):\n def Fan_page(self):\n print(str(input(\"Enter your name\")))\n print(str(input(\"Which club do you support?\")))\n print(\"WELCOME TO TSG WALKERS FC!\")\n\n \n \n \na = FC_Cirok()\na.Fan_page()\nb = Madiba_FC()\nb.Fan_page()\nc = Blue_Jay_FC()\nc.Fan_page()\nd = TSG_Walkers()\nd.Fan_page()\n\n\n \n ", "Enter your name Chima\n Chima\nWhich club do you support? Cirok\n Cirok\nWELCOME TO CIROK FC!\nEnter your name Toju\n Toju\nWhich club do you support? Madiba\n Madiba\nWELCOME TO MADIBA FC!\nEnter your name Daniel\n Daniel\nWhich club do you support? Bluejays\n Bluejays\nWELCOME TO THE BLUES!\nEnter your name Murewa\n Murewa\nWhich club do you support? TSG\n TSG\nWELCOME TO TSG WALKERS FC!\n" ] ], [ [ "## Class Project II", "_____no_output_____" ], [ "The Service Unit of PAU has contacted you to develop a program to manage some of the External Food Vendors. With your knowledge in python OOP develop a program to manage the PAU External Food Vendors. The program receives as input the vendor of interest and display the menu of the interested vendor. The External vendors are Faith hostel, Cooperative Hostel, and Student Center. Find below the menus:\n\n<table><tr><td>\n<table style=\"background-color:#47b5ff\">\n <tr><th colspan='2'>Cooperative Cafeteria</th></tr>\n <tr><th>Main Meal</th><th>Price (N)</th></tr>\n <tr><td>Jollof Rice and Stew</td><td>200</td></tr>\n <tr><td>White Rice and Stew</td><td>200</td></tr>\n <tr><td>Fried Rice</td><td>200</td></tr>\n <tr><td>Salad</td><td>100</td></tr>\n <tr><td>Platain</td><td>100</td></tr>\n</table>\n </td><td>\n<table style=\"background-color:pink\">\n <tr><th colspan='2'>Faith Hostel Cafeteria</th></tr>\n <tr><th>Main Meal</th><th>Price (N)</th></tr>\n <tr><td>Fried Rice</td><td>400</td></tr>\n <tr><td>White Rice and Stew</td><td>400</td></tr>\n <tr><td>Jollof Rice</td><td>400</td></tr>\n <tr><td>Beans</td><td>200</td></tr>\n <tr><td>Chicken</td><td>1000</td></tr>\n</table>\n </td><td>\n <table style=\"background-color:#fcf96c\">\n <tr><th colspan='2'>Student Centre Cafeteria</th></tr>\n <tr><th>Main Meal</th><th>Price (N)</th></tr>\n <tr><td>Chicken Fried Rice</td><td>800</td></tr>\n <tr><td>Pomo Sauce</td><td>300</td></tr>\n <tr><td>Spaghetti Jollof</td><td>500</td></tr>\n <tr><td>Amala/Ewedu</td><td>500</td></tr>\n <tr><td>Semo with Eforiro Soup</td><td>500</td></tr>\n</table>\n </td></tr>\n<table>\n \n<p><b>Hints:</b></p>\n <ul><li>The abstract base class is called <b>External_Vendors()</b>.</li><li>\n The abstract method is called <b>menu()</b>.</li><li>\nThe subclasses (the different vendors) will inherit the abstract base class.</li><li>\n Each subclass will have a normal method called <b>menu()</b>.</li></ul>\n \n ", "_____no_output_____" ] ], [ [ "from abc import ABC, abstractmethod\nclass External_Vendors(ABC):\n \n @abstractmethod\n def menu(self):\n pass\n\nclass Cooperative_cafeteria(External_Vendors):\n def menu(self):\n print(str(input(\"Which external vendor would you prefer?\")))\n print(\"Menu ; Jollof Rice and Stew, White Rice and Stew, Fried Rice, Salad, Platain\")\n\nclass Faith_Hostel_Cafeteria(External_Vendors):\n def menu(self):\n print(str(input(\"Which external vendor would you prefer?\")))\n print(\"Menu ; Jollof Rice , White Rice and Stew, Fried Rice, Beans, Chicken\")\n\nclass Student_centre_cafeteria(External_Vendors):\n def menu(self):\n print(str(input(\"Which external vendor would you prefer?\")))\n print(\"Menu ; Pomo sauce, Chicken Fried Rice, Spaghetti Jollof, Amala/Ewedu, Semo with Efo riro soup\")\n \na = Cooperative_cafeteria()\na.menu()\nb = Faith_Hostel_Cafeteria()\nb.menu()\nc = Student_centre_cafeteria()\nc.menu()\n ", "Which external vendor would you prefer? Cooperative\n Cooperative\nMenu ; Jollof Rice and Stew, White Rice and Stew, Fried Rice, Salad, Platain\nWhich external vendor would you prefer? Faith\n Faith\nMenu ; Jollof Rice , White Rice and Stew, Fried Rice, Beans, Chicken\nWhich external vendor would you prefer? Students Centre\n Students Centre\nMenu ; Pomo sauce, Chicken Fried Rice, Spaghetti Jollof, Amala/Ewedu, Semo with Efo riro soup\n" ] ] ]

[ "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code" ]

[ [ "markdown", "markdown", "markdown", "markdown", "markdown", "markdown", "markdown", "markdown", "markdown", "markdown", "markdown", "markdown", "markdown" ], [ "code" ], [ "markdown" ], [ "code" ], [ "markdown" ], [ "code" ], [ "markdown" ], [ "code" ], [ "markdown" ], [ "code" ], [ "markdown" ], [ "code" ], [ "markdown" ], [ "code" ], [ "markdown", "markdown" ], [ "code" ], [ "markdown" ], [ "code" ], [ "markdown", "markdown" ], [ "code" ], [ "markdown", "markdown" ], [ "code" ] ]

[ "Apache-2.0" ]

[ "Apache-2.0" ]

[ "Apache-2.0" ]

[ [ [ "# Boston Housing KNN", "_____no_output_____" ] ], [ [ "import sys\nsys.path.append(\"..\")\nfrom pyspark.sql.types import BooleanType\nfrom pyspark.ml.feature import StandardScaler, VectorAssembler, BucketedRandomProjectionLSH\nfrom pyspark.ml.classification import LinearSVC\nfrom pyspark.sql import Row\nfrom pyspark.sql.session import SparkSession\nfrom pyspark.sql.functions import desc, expr\nfrom pyspark.ml.evaluation import BinaryClassificationEvaluator\nfrom helpers.path_translation import translate_to_file_string", "_____no_output_____" ], [ "inputFile = translate_to_file_string(\"../data/Boston_Housing_Data.csv\")", "_____no_output_____" ] ], [ [ "Spark session creation ", "_____no_output_____" ] ], [ [ "spark = (SparkSession\n .builder\n .appName(\"BostonHousingKNN\")\n .getOrCreate())", "_____no_output_____" ] ], [ [ "DataFrame creation using an ifered Schema ", "_____no_output_____" ] ], [ [ "df = spark.read.option(\"header\", \"true\") \\\n .option(\"inferSchema\", \"true\") \\\n .option(\"delimiter\", \";\") \\\n .csv(inputFile) \\\n .withColumn(\"CATBOOL\", expr(\"CAT\").cast(BooleanType()))\nprint(df.printSchema())", "_____no_output_____" ] ], [ [ "Prepare training and test data.", "_____no_output_____" ] ], [ [ "featureCols = df.columns.copy()\nfeatureCols.remove(\"MEDV\")\nfeatureCols.remove(\"CAT\")\nfeatureCols.remove(\"CATBOOL\") \nprint(featureCols)\n\nassembler = VectorAssembler(outputCol=\"features\", inputCols=featureCols)\nscaler = StandardScaler(inputCol=\"features\", outputCol=\"scaledFeatures\",\n withStd=True, withMean=False)", "_____no_output_____" ], [ "labledPointDataSet = assembler.transform(df)\nscaledDataSet = scaler.fit(labledPointDataSet).transform(labledPointDataSet)\nsplits = scaledDataSet.randomSplit([0.9, 0.1 ], 12345)\ntraining = splits[0]\ntest = splits[1]", "_____no_output_____" ] ], [ [ "LHS Euclidean Distance", "_____no_output_____" ] ], [ [ "# TODO optimize the params to minimize the test error\n# TODO try the MinHashLSH too\nlhsED = BucketedRandomProjectionLSH(inputCol=\"scaledFeatures\", outputCol=\"hashes\", bucketLength =2.0, numHashTables=3)", "_____no_output_____" ] ], [ [ "Train the model ", "_____no_output_____" ] ], [ [ "modelED = lhsED.fit(training)", "_____no_output_____" ] ], [ [ "Test the model", "_____no_output_____" ] ], [ [ "resultList = []\n# The Nearest neighbor testing\n# TODO add other aggregation methods \nfor row in test.collect() :\n neighbors = modelED.approxNearestNeighbors(training, row.scaledFeatures, 5)\n grouped = neighbors.groupBy(df.CAT).count()\n if grouped.count() > 0 :\n result = grouped.orderBy(desc(\"count\")).first().CAT\n newRow = Row(CAT=row.CAT, scaledFeatures=row.scaledFeatures, prediction=float (result))\n resultList.append(newRow)\t\n\npredictions = SparkSession.builder.getOrCreate().createDataFrame(resultList)\npredictions.show()", "_____no_output_____" ], [ "evaluator = BinaryClassificationEvaluator(labelCol=\"CAT\",rawPredictionCol=\"prediction\", metricName=\"areaUnderROC\")\naccuracy = evaluator.evaluate(predictions)\nprint(\"Test Error\",(1.0 - accuracy))", "_____no_output_____" ], [ "spark.stop()", "_____no_output_____" ] ] ]

[ "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code" ]

[ [ "markdown" ], [ "code", "code" ], [ "markdown" ], [ "code" ], [ "markdown" ], [ "code" ], [ "markdown" ], [ "code", "code" ], [ "markdown" ], [ "code" ], [ "markdown" ], [ "code" ], [ "markdown" ], [ "code", "code", "code" ] ]

[ "MIT" ]

[ "MIT" ]

[ "MIT" ]

[ [ [ "import turicreate", "_____no_output_____" ] ], [ [ "# Load some text data - from wikipedia, pages on people", "_____no_output_____" ] ], [ [ "people = turicreate.SFrame('people_wiki.sframe')", "_____no_output_____" ], [ "people", "_____no_output_____" ] ], [ [ "# Explore the dataset and checkout the text it contains", "_____no_output_____" ] ], [ [ "obama = people[people['name'] == 'Barack Obama']", "_____no_output_____" ], [ "obama", "_____no_output_____" ], [ "obama['text']", "_____no_output_____" ] ], [ [ "## Explore the entry for actor George Clooney", "_____no_output_____" ] ], [ [ "clooney = people[people['name'] == 'George Clooney']\nclooney['text']", "_____no_output_____" ] ], [ [ "# Word counts for Obama acticle", "_____no_output_____" ] ], [ [ "obama['word_count'] = turicreate.text_analytics.count_words(obama['text'])", "_____no_output_____" ], [ "obama['word_count']", "_____no_output_____" ] ], [ [ "## Sort the word counts for the Obama article", "_____no_output_____" ] ], [ [ "obama.stack('word_count',new_column_name=['word','count'])", "_____no_output_____" ], [ "obama_word_count_table = obama[['word_count']].stack('word_count', new_column_name = ['word','count'])", "_____no_output_____" ], [ "obama_word_count_table", "_____no_output_____" ], [ "obama_word_count_table.sort('count',ascending=False)", "_____no_output_____" ] ], [ [ "## Compute TF-IDF for the corpus", "_____no_output_____" ] ], [ [ "people['word_count'] = turicreate.text_analytics.count_words(people['text'])", "_____no_output_____" ], [ "people", "_____no_output_____" ], [ "people['tfidf'] = turicreate.text_analytics.tf_idf(people['text'])", "_____no_output_____" ], [ "people", "_____no_output_____" ] ], [ [ "## Examine the TF-IDF for the Obama article", "_____no_output_____" ] ], [ [ "obama = people[people['name'] == 'Barack Obama']\nobama[['tfidf']].stack('tfidf',new_column_name=['word','tfidf']).sort('tfidf',ascending=False)", "_____no_output_____" ] ], [ [ "# Manually compute distances between a few people", "_____no_output_____" ] ], [ [ "clinton = people[people['name'] == 'Bill Clinton']", "_____no_output_____" ], [ "beckham = people[people['name'] == 'David Beckham']", "_____no_output_____" ] ], [ [ "## Is Obama closer to Clinton than to Beckham?", "_____no_output_____" ] ], [ [ "turicreate.distances.cosine(obama['tfidf'][0],clinton['tfidf'][0])", "_____no_output_____" ], [ "# The smaller the cosine value, the more relevant it is.", "_____no_output_____" ], [ "turicreate.distances.cosine(obama['tfidf'][0],beckham['tfidf'][0])", "_____no_output_____" ] ], [ [ "# Build the nearest neighbor model for document retrieval", "_____no_output_____" ] ], [ [ "knn_model = turicreate.nearest_neighbors.create(people,features=['tfidf'],label='name')", "_____no_output_____" ] ], [ [ "# Applying the nearest-neighbors model for retrieval", "_____no_output_____" ], [ "## Who is closest to Obama?", "_____no_output_____" ] ], [ [ "knn_model.query(obama)", "_____no_output_____" ] ], [ [ "# Other examples of document retrieval", "_____no_output_____" ] ], [ [ "taylor_swift = people[people['name'] == 'Taylor Swift']", "_____no_output_____" ], [ "knn_model.query(taylor_swift)", "_____no_output_____" ], [ "jolie = people[people['name'] == 'Angelina Jolie']", "_____no_output_____" ], [ "knn_model.query(jolie)", "_____no_output_____" ], [ "arnold = people[people['name'] == 'Arnold Schwarzenegger']", "_____no_output_____" ], [ "knn_model.query(arnold)", "_____no_output_____" ] ] ]

[ "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code" ]

[ [ "code" ], [ "markdown" ], [ "code", "code" ], [ "markdown" ], [ "code", "code", "code" ], [ "markdown" ], [ "code" ], [ "markdown" ], [ "code", "code" ], [ "markdown" ], [ "code", "code", "code", "code" ], [ "markdown" ], [ "code", "code", "code", "code" ], [ "markdown" ], [ "code" ], [ "markdown" ], [ "code", "code" ], [ "markdown" ], [ "code", "code", "code" ], [ "markdown" ], [ "code" ], [ "markdown", "markdown" ], [ "code" ], [ "markdown" ], [ "code", "code", "code", "code", "code", "code" ] ]

[ "MIT" ]

[ "MIT" ]

[ "MIT" ]

[ [ [ "# Chapter 3\n\nThis is the third in a series of notebooks related to astronomy data.\n\nAs a running example, we are replicating parts of the analysis in a recent paper, \"[Off the beaten path: Gaia reveals GD-1 stars outside of the main stream](https://arxiv.org/abs/1805.00425)\" by Adrian M. Price-Whelan and Ana Bonaca.\n\nIn the first lesson, we wrote ADQL queries and used them to select and download data from the Gaia server.\n\nIn the second lesson, we wrote a query to select stars from the region of the sky where we expect GD-1 to be, and saved the results in a FITS file.\n\nNow we'll read that data back and implement the next step in the analysis, identifying stars with the proper motion we expect for GD-1.", "_____no_output_____" ], [ "## Outline\n\nHere are the steps in this lesson:\n\n1. We'll read back the results from the previous lesson, which we saved in a FITS file.\n\n2. Then we'll transform the coordinates and proper motion data from ICRS back to the coordinate frame of GD-1.\n\n3. We'll put those results into a Pandas `DataFrame`, which we'll use to select stars near the centerline of GD-1.\n\n4. Plotting the proper motion of those stars, we'll identify a region of proper motion for stars that are likely to be in GD-1.\n\n5. Finally, we'll select and plot the stars whose proper motion is in that region.\n\nAfter completing this lesson, you should be able to\n\n* Select rows and columns from an Astropy `Table`.\n\n* Use Matplotlib to make a scatter plot.\n\n* Use Gala to transform coordinates.\n\n* Make a Pandas `DataFrame` and use a Boolean `Series` to select rows.\n\n* Save a `DataFrame` in an HDF5 file.\n", "_____no_output_____" ], [ "## Installing libraries\n\nIf you are running this notebook on Colab, you can run the following cell to install Astroquery and the other libraries we'll use.\n\nIf you are running this notebook on your own computer, you might have to install these libraries yourself. See the instructions in the preface.", "_____no_output_____" ] ], [ [ "# If we're running on Colab, install libraries\n\nimport sys\nIN_COLAB = 'google.colab' in sys.modules\n\nif IN_COLAB:\n !pip install astroquery astro-gala pyia python-wget", "_____no_output_____" ] ], [ [ "## Reload the data\n\nIn the previous lesson, we ran a query on the Gaia server and downloaded data for roughly 100,000 stars. We saved the data in a FITS file so that now, picking up where we left off, we can read the data from a local file rather than running the query again.\n\nIf you ran the previous lesson successfully, you should already have a file called `gd1_results.fits` that contains the data we downloaded.\n\nIf not, you can run the following cell, which downloads the data from our repository.", "_____no_output_____" ] ], [ [ "import os\nfrom wget import download\n\nfilename = 'gd1_results.fits'\npath = 'https://github.com/AllenDowney/AstronomicalData/raw/main/data/'\n\nif not os.path.exists(filename):\n print(download(path+filename))", "_____no_output_____" ] ], [ [ "Now here's how we can read the data from the file back into an Astropy `Table`:", "_____no_output_____" ] ], [ [ "from astropy.table import Table\n\nresults = Table.read(filename)", "_____no_output_____" ] ], [ [ "The result is an Astropy `Table`.\n\nWe can use `info` to refresh our memory of the contents.", "_____no_output_____" ] ], [ [ "results.info", "_____no_output_____" ] ], [ [ "## Selecting rows and columns\n\nIn this section we'll see operations for selecting columns and rows from an Astropy `Table`. You can find more information about these operations in the [Astropy documentation](https://docs.astropy.org/en/stable/table/access_table.html).\n\nWe can get the names of the columns like this:", "_____no_output_____" ] ], [ [ "results.colnames", "_____no_output_____" ] ], [ [ "And select an individual column like this:", "_____no_output_____" ] ], [ [ "results['ra']", "_____no_output_____" ] ], [ [ "The result is a `Column` object that contains the data, and also the data type, units, and name of the column.", "_____no_output_____" ] ], [ [ "type(results['ra'])", "_____no_output_____" ] ], [ [ "The rows in the `Table` are numbered from 0 to `n-1`, where `n` is the number of rows. We can select the first row like this:", "_____no_output_____" ] ], [ [ "results[0]", "_____no_output_____" ] ], [ [ "As you might have guessed, the result is a `Row` object.", "_____no_output_____" ] ], [ [ "type(results[0])", "_____no_output_____" ] ], [ [ "Notice that the bracket operator selects both columns and rows. You might wonder how it knows which to select.\n\nIf the expression in brackets is a string, it selects a column; if the expression is an integer, it selects a row.\n\nIf you apply the bracket operator twice, you can select a column and then an element from the column.", "_____no_output_____" ] ], [ [ "results['ra'][0]", "_____no_output_____" ] ], [ [ "Or you can select a row and then an element from the row.", "_____no_output_____" ] ], [ [ "results[0]['ra']", "_____no_output_____" ] ], [ [ "You get the same result either way.", "_____no_output_____" ], [ "## Scatter plot\n\nTo see what the results look like, we'll use a scatter plot. The library we'll use is [Matplotlib](https://matplotlib.org/), which is the most widely-used plotting library for Python.\n\nThe Matplotlib interface is based on MATLAB (hence the name), so if you know MATLAB, some of it will be familiar.\n\nWe'll import like this.", "_____no_output_____" ] ], [ [ "import matplotlib.pyplot as plt", "_____no_output_____" ] ], [ [ "Pyplot part of the Matplotlib library. It is conventional to import it using the shortened name `plt`.\n\nPyplot provides two functions that can make scatterplots, [plt.scatter](https://matplotlib.org/3.3.0/api/_as_gen/matplotlib.pyplot.scatter.html) and [plt.plot](https://matplotlib.org/api/_as_gen/matplotlib.pyplot.plot.html).\n\n* `scatter` is more versatile; for example, you can make every point in a scatter plot a different color.\n\n* `plot` is more limited, but for simple cases, it can be substantially faster. \n\nJake Vanderplas explains these differences in [The Python Data Science Handbook](https://jakevdp.github.io/PythonDataScienceHandbook/04.02-simple-scatter-plots.html)\n\nSince we are plotting more than 100,000 points and they are all the same size and color, we'll use `plot`.\n\nHere's a scatter plot with right ascension on the x-axis and declination on the y-axis, both ICRS coordinates in degrees.", "_____no_output_____" ] ], [ [ "x = results['ra']\ny = results['dec']\nplt.plot(x, y, 'ko')\n\nplt.xlabel('ra (degree ICRS)')\nplt.ylabel('dec (degree ICRS)');", "_____no_output_____" ] ], [ [ "The arguments to `plt.plot` are `x`, `y`, and a string that specifies the style. In this case, the letters `ko` indicate that we want a black, round marker (`k` is for black because `b` is for blue).\n\nThe functions `xlabel` and `ylabel` put labels on the axes.\n\nThis scatter plot has a problem. It is \"[overplotted](https://python-graph-gallery.com/134-how-to-avoid-overplotting-with-python/)\", which means that there are so many overlapping points, we can't distinguish between high and low density areas.\n\nTo fix this, we can provide optional arguments to control the size and transparency of the points.", "_____no_output_____" ], [ "## Exercise\n\nIn the call to `plt.plot`, add the keyword argument `markersize=0.1` to make the markers smaller.\n\nThen add the argument `alpha=0.1` to make the markers nearly transparent.\n\nAdjust these arguments until you think the figure shows the data most clearly.\n\nNote: Once you have made these changes, you might notice that the figure shows stripes with lower density of stars. These stripes are caused by the way Gaia scans the sky, which [you can read about here](https://www.cosmos.esa.int/web/gaia/scanning-law). The dataset we are using, [Gaia Data Release 2](https://www.cosmos.esa.int/web/gaia/dr2), covers 22 months of observations; during this time, some parts of the sky were scanned more than others.", "_____no_output_____" ], [ "## Transform back\n\nRemember that we selected data from a rectangle of coordinates in the `GD1Koposov10` frame, then transformed them to ICRS when we constructed the query.\nThe coordinates in `results` are in ICRS.\n\nTo plot them, we will transform them back to the `GD1Koposov10` frame; that way, the axes of the figure are aligned with the GD-1, which will make it easy to select stars near the centerline of the stream.\n\nTo do that, we'll put the results into a `GaiaData` object, provided by the [pyia library](https://pyia.readthedocs.io/en/latest/api/pyia.GaiaData.html).\n\nTODO: Do we need pyia, or could we do this with astropy and gala?", "_____no_output_____" ] ], [ [ "from pyia import GaiaData\n\ngaia_data = GaiaData(results)\ntype(gaia_data)", "_____no_output_____" ] ], [ [ "Now we can extract sky coordinates from the `GaiaData` object, like this:", "_____no_output_____" ] ], [ [ "import astropy.units as u\n\nskycoord = gaia_data.get_skycoord(\n distance=8*u.kpc, \n radial_velocity=0*u.km/u.s)", "_____no_output_____" ] ], [ [ "We provide `distance` and `radial_velocity` to prepare the data for reflex correction, which we explain below.", "_____no_output_____" ] ], [ [ "type(skycoord)", "_____no_output_____" ] ], [ [ "The result is an Astropy `SkyCoord` object ([documentation here](https://docs.astropy.org/en/stable/api/astropy.coordinates.SkyCoord.html#astropy.coordinates.SkyCoord)), which provides `transform_to`, so we can transform the coordinates to other frames.", "_____no_output_____" ] ], [ [ "import gala.coordinates as gc\n\ntransformed = skycoord.transform_to(gc.GD1Koposov10)\ntype(transformed)", "_____no_output_____" ] ], [ [ "The result is another `SkyCoord` object, now in the `GD1Koposov10` frame.", "_____no_output_____" ], [ "The next step is to correct the proper motion measurements from Gaia for reflex due to the motion of our solar system around the Galactic center.\n\nWhen we created `skycoord`, we provided `distance` and `radial_velocity` as arguments, which means we ignore the measurements provided by Gaia and replace them with these fixed values.\n\nThat might seem like a strange thing to do, but here's the motivation:\n\n* Because the stars in GD-1 are so far away, the distance estimates we get from Gaia, which are based on parallax, are not very precise. So we replace them with our current best estimate of the mean distance to GD-1, about 8 kpc. See [Koposov, Rix, and Hogg, 2010](https://ui.adsabs.harvard.edu/abs/2010ApJ...712..260K/abstract).\n\n* For the other stars in the table, this distance estimate will be inaccurate, so reflex correction will not be correct. But that should have only a small effect on our ability to identify stars with the proper motion we expect for GD-1.\n\n* The measurement of radial velocity has no effect on the correction for proper motion; the value we provide is arbitrary, but we have to provide a value to avoid errors in the reflex correction calculation.\n\nWe are grateful to Adrian Price-Whelen for his help explaining this step in the analysis.", "_____no_output_____" ], [ "With this preparation, we can use `reflex_correct` from Gala ([documentation here](https://gala-astro.readthedocs.io/en/latest/api/gala.coordinates.reflex_correct.html)) to correct for solar reflex motion.", "_____no_output_____" ] ], [ [ "gd1_coord = gc.reflex_correct(transformed)\n\ntype(gd1_coord)", "_____no_output_____" ] ], [ [ "The result is a `SkyCoord` object that contains \n\n* The transformed coordinates as attributes named `phi1` and `phi2`, which represent right ascension and declination in the `GD1Koposov10` frame.\n\n* The transformed and corrected proper motions as `pm_phi1_cosphi2` and `pm_phi2`.\n\nWe can select the coordinates like this:", "_____no_output_____" ] ], [ [ "phi1 = gd1_coord.phi1\nphi2 = gd1_coord.phi2", "_____no_output_____" ] ], [ [ "And plot them like this:", "_____no_output_____" ] ], [ [ "plt.plot(phi1, phi2, 'ko', markersize=0.1, alpha=0.2)\n\nplt.xlabel('ra (degree GD1)')\nplt.ylabel('dec (degree GD1)');", "_____no_output_____" ] ], [ [ "Remember that we started with a rectangle in GD-1 coordinates. When transformed to ICRS, it's a non-rectangular polygon. Now that we have transformed back to GD-1 coordinates, it's a rectangle again.", "_____no_output_____" ], [ "## Pandas DataFrame\n\nAt this point we have three objects containing different subsets of the data.", "_____no_output_____" ] ], [ [ "type(results)", "_____no_output_____" ], [ "type(gaia_data)", "_____no_output_____" ], [ "type(gd1_coord)", "_____no_output_____" ] ], [ [ "On one hand, this makes sense, since each object provides different capabilities. But working with three different object types can be awkward.\n\nIt will be more convenient to choose one object and get all of the data into it. We'll use a Pandas DataFrame, for two reasons:\n\n1. It provides capabilities that are pretty much a superset of the other data structures, so it's the all-in-one solution.\n\n2. Pandas is a general-purpose tool that is useful in many domains, especially data science. If you are going to develop expertise in one tool, Pandas is a good choice.\n\nHowever, compared to an Astropy `Table`, Pandas has one big drawback: it does not keep the metadata associated with the table, including the units for the columns.", "_____no_output_____" ], [ "It's easy to convert a `Table` to a Pandas `DataFrame`.", "_____no_output_____" ] ], [ [ "import pandas as pd\n\ndf = results.to_pandas()\ndf.shape", "_____no_output_____" ] ], [ [ "`DataFrame` provides `shape`, which shows the number of rows and columns.\n\nIt also provides `head`, which displays the first few rows. It is useful for spot-checking large results as you go along.", "_____no_output_____" ] ], [ [ "df.head()", "_____no_output_____" ] ], [ [ "Python detail: `shape` is an attribute, so we can display it's value without calling it as a function; `head` is a function, so we need the parentheses.", "_____no_output_____" ], [ "Now we can extract the columns we want from `gd1_coord` and add them as columns in the `DataFrame`. `phi1` and `phi2` contain the transformed coordinates.", "_____no_output_____" ] ], [ [ "df['phi1'] = gd1_coord.phi1\ndf['phi2'] = gd1_coord.phi2\ndf.shape", "_____no_output_____" ] ], [ [ "`pm_phi1_cosphi2` and `pm_phi2` contain the components of proper motion in the transformed frame.", "_____no_output_____" ] ], [ [ "df['pm_phi1'] = gd1_coord.pm_phi1_cosphi2\ndf['pm_phi2'] = gd1_coord.pm_phi2\ndf.shape", "_____no_output_____" ] ], [ [ "**Detail:** If you notice that `SkyCoord` has an attribute called `proper_motion`, you might wonder why we are not using it.\n\nWe could have: `proper_motion` contains the same data as `pm_phi1_cosphi2` and `pm_phi2`, but in a different format.", "_____no_output_____" ], [ "## Plot proper motion\n\nNow we are ready to replicate one of the panels in Figure 1 of the Price-Whelan and Bonaca paper, the one that shows the components of proper motion as a scatter plot:\n\n<img width=\"300\" src=\"https://github.com/datacarpentry/astronomy-python/raw/gh-pages/fig/gd1-1.png\">\n\nIn this figure, the shaded area is a high-density region of stars with the proper motion we expect for stars in GD-1. \n\n* Due to the nature of tidal streams, we expect the proper motion for most stars to be along the axis of the stream; that is, we expect motion in the direction of `phi2` to be near 0.\n\n* In the direction of `phi1`, we don't have a prior expectation for proper motion, except that it should form a cluster at a non-zero value. \n\nTo locate this cluster, we'll select stars near the centerline of GD-1 and plot their proper motion.", "_____no_output_____" ], [ "## Selecting the centerline\n\nAs we can see in the following figure, many stars in GD-1 are less than 1 degree of declination from the line `phi2=0`.\n\n<img src=\"https://github.com/datacarpentry/astronomy-python/raw/gh-pages/fig/gd1-4.png\">\n\nIf we select stars near this line, they are more likely to be in GD-1.\n\nWe'll start by selecting the `phi2` column from the `DataFrame`:", "_____no_output_____" ] ], [ [ "phi2 = df['phi2']\ntype(phi2)", "_____no_output_____" ] ], [ [ "The result is a `Series`, which is the structure Pandas uses to represent columns.\n\nWe can use a comparison operator, `>`, to compare the values in a `Series` to a constant.", "_____no_output_____" ] ], [ [ "phi2_min = -1.0 * u.deg\nphi2_max = 1.0 * u.deg\n\nmask = (df['phi2'] > phi2_min)\ntype(mask)", "_____no_output_____" ], [ "mask.dtype", "_____no_output_____" ] ], [ [ "The result is a `Series` of Boolean values, that is, `True` and `False`. ", "_____no_output_____" ] ], [ [ "mask.head()", "_____no_output_____" ] ], [ [ "A Boolean `Series` is sometimes called a \"mask\" because we can use it to mask out some of the rows in a `DataFrame` and select the rest, like this:", "_____no_output_____" ] ], [ [ "subset = df[mask]\ntype(subset)", "_____no_output_____" ] ], [ [ "`subset` is a `DataFrame` that contains only the rows from `df` that correspond to `True` values in `mask`.\n\nThe previous mask selects all stars where `phi2` exceeds `phi2_min`; now we'll select stars where `phi2` falls between `phi2_min` and `phi2_max`.", "_____no_output_____" ] ], [ [ "phi_mask = ((df['phi2'] > phi2_min) & \n (df['phi2'] < phi2_max))", "_____no_output_____" ] ], [ [ "The `&` operator computes \"logical AND\", which means the result is true where elements from both Boolean `Series` are true.\n\nThe sum of a Boolean `Series` is the number of `True` values, so we can use `sum` to see how many stars are in the selected region.", "_____no_output_____" ] ], [ [ "phi_mask.sum()", "_____no_output_____" ] ], [ [ "And we can use `phi1_mask` to select stars near the centerline, which are more likely to be in GD-1.", "_____no_output_____" ] ], [ [ "centerline = df[phi_mask]\nlen(centerline)", "_____no_output_____" ] ], [ [ "Here's a scatter plot of proper motion for the selected stars.", "_____no_output_____" ] ], [ [ "pm1 = centerline['pm_phi1']\npm2 = centerline['pm_phi2']\n\nplt.plot(pm1, pm2, 'ko', markersize=0.1, alpha=0.1)\n \nplt.xlabel('Proper motion phi1 (GD1 frame)')\nplt.ylabel('Proper motion phi2 (GD1 frame)');", "_____no_output_____" ] ], [ [ "Looking at these results, we see a large cluster around (0, 0), and a smaller cluster near (0, -10).\n\nWe can use `xlim` and `ylim` to set the limits on the axes and zoom in on the region near (0, 0).", "_____no_output_____" ] ], [ [ "pm1 = centerline['pm_phi1']\npm2 = centerline['pm_phi2']\n\nplt.plot(pm1, pm2, 'ko', markersize=0.3, alpha=0.3)\n \nplt.xlabel('Proper motion phi1 (GD1 frame)')\nplt.ylabel('Proper motion phi2 (GD1 frame)')\n\nplt.xlim(-12, 8)\nplt.ylim(-10, 10);", "_____no_output_____" ] ], [ [ "Now we can see the smaller cluster more clearly.\n\nYou might notice that our figure is less dense than the one in the paper. That's because we started with a set of stars from a relatively small region. The figure in the paper is based on a region about 10 times bigger.\n\nIn the next lesson we'll go back and select stars from a larger region. But first we'll use the proper motion data to identify stars likely to be in GD-1.", "_____no_output_____" ], [ "## Filtering based on proper motion\n\nThe next step is to select stars in the \"overdense\" region of proper motion, which are candidates to be in GD-1.\n\nIn the original paper, Price-Whelan and Bonaca used a polygon to cover this region, as shown in this figure.\n\n<img width=\"300\" src=\"https://github.com/datacarpentry/astronomy-python/raw/gh-pages/fig/gd1-1.png\">\n\nWe'll use a simple rectangle for now, but in a later lesson we'll see how to select a polygonal region as well.\n\nHere are bounds on proper motion we chose by eye,", "_____no_output_____" ] ], [ [ "pm1_min = -8.9\npm1_max = -6.9\npm2_min = -2.2\npm2_max = 1.0", "_____no_output_____" ] ], [ [ "To draw these bounds, we'll make two lists containing the coordinates of the corners of the rectangle.", "_____no_output_____" ] ], [ [ "pm1_rect = [pm1_min, pm1_min, pm1_max, pm1_max, pm1_min] * u.mas/u.yr\npm2_rect = [pm2_min, pm2_max, pm2_max, pm2_min, pm2_min] * u.mas/u.yr", "_____no_output_____" ] ], [ [ "Here's what the plot looks like with the bounds we chose.", "_____no_output_____" ] ], [ [ "plt.plot(pm1, pm2, 'ko', markersize=0.3, alpha=0.3)\nplt.plot(pm1_rect, pm2_rect, '-')\n \nplt.xlabel('Proper motion phi1 (GD1 frame)')\nplt.ylabel('Proper motion phi2 (GD1 frame)')\n\nplt.xlim(-12, 8)\nplt.ylim(-10, 10);", "_____no_output_____" ] ], [ [ "To select rows that fall within these bounds, we'll use the following function, which uses Pandas operators to make a mask that selects rows where `series` falls between `low` and `high`.", "_____no_output_____" ] ], [ [ "def between(series, low, high):\n \"\"\"Make a Boolean Series.\n \n series: Pandas Series\n low: lower bound\n high: upper bound\n \n returns: Boolean Series\n \"\"\"\n return (series > low) & (series < high)", "_____no_output_____" ] ], [ [ "The following mask select stars with proper motion in the region we chose.", "_____no_output_____" ] ], [ [ "pm_mask = (between(df['pm_phi1'], pm1_min, pm1_max) & \n between(df['pm_phi2'], pm2_min, pm2_max))", "_____no_output_____" ] ], [ [ "Again, the sum of a Boolean series is the number of `True` values.", "_____no_output_____" ] ], [ [ "pm_mask.sum()", "_____no_output_____" ] ], [ [ "Now we can use this mask to select rows from `df`.", "_____no_output_____" ] ], [ [ "selected = df[pm_mask]\nlen(selected)", "_____no_output_____" ] ], [ [ "These are the stars we think are likely to be in GD-1. Let's see what they look like, plotting their coordinates (not their proper motion).", "_____no_output_____" ] ], [ [ "phi1 = selected['phi1']\nphi2 = selected['phi2']\n\nplt.plot(phi1, phi2, 'ko', markersize=0.5, alpha=0.5)\n\nplt.xlabel('ra (degree GD1)')\nplt.ylabel('dec (degree GD1)');", "_____no_output_____" ] ], [ [ "Now that's starting to look like a tidal stream!", "_____no_output_____" ], [ "## Saving the DataFrame\n\nAt this point we have run a successful query and cleaned up the results; this is a good time to save the data.\n\nTo save a Pandas `DataFrame`, one option is to convert it to an Astropy `Table`, like this:", "_____no_output_____" ] ], [ [ "selected_table = Table.from_pandas(selected)\ntype(selected_table)", "_____no_output_____" ] ], [ [ "Then we could write the `Table` to a FITS file, as we did in the previous lesson. \n\nBut Pandas provides functions to write DataFrames in other formats; to see what they are [find the functions here that begin with `to_`](https://pandas.pydata.org/pandas-docs/stable/reference/api/pandas.DataFrame.html).\n\nOne of the best options is HDF5, which is Version 5 of [Hierarchical Data Format](https://en.wikipedia.org/wiki/Hierarchical_Data_Format).\n\nHDF5 is a binary format, so files are small and fast to read and write (like FITS, but unlike XML).\n\nAn HDF5 file is similar to an SQL database in the sense that it can contain more than one table, although in HDF5 vocabulary, a table is called a Dataset. ([Multi-extension FITS files](https://www.stsci.edu/itt/review/dhb_2011/Intro/intro_ch23.html) can also contain more than one table.)\n\nAnd HDF5 stores the metadata associated with the table, including column names, row labels, and data types (like FITS).\n\nFinally, HDF5 is a cross-language standard, so if you write an HDF5 file with Pandas, you can read it back with many other software tools (more than FITS).", "_____no_output_____" ], [ "We can write a Pandas `DataFrame` to an HDF5 file like this:", "_____no_output_____" ] ], [ [ "filename = 'gd1_dataframe.hdf5'\n\ndf.to_hdf(filename, 'df', mode='w')", "_____no_output_____" ] ], [ [ "Because an HDF5 file can contain more than one Dataset, we have to provide a name, or \"key\", that identifies the Dataset in the file.\n\nWe could use any string as the key, but in this example I use the variable name `df`.", "_____no_output_____" ], [ "## Exercise \n\nWe're going to need `centerline` and `selected` later as well. Write a line or two of code to add it as a second Dataset in the HDF5 file.", "_____no_output_____" ] ], [ [ "# Solution\n\ncenterline.to_hdf(filename, 'centerline')\nselected.to_hdf(filename, 'selected')", "_____no_output_____" ] ], [ [ "**Detail:** Reading and writing HDF5 tables requires a library called `PyTables` that is not always installed with Pandas. You can install it with pip like this:\n\n```\npip install tables\n```\n\nIf you install it using Conda, the name of the package is `pytables`.\n\n```\nconda install pytables\n```", "_____no_output_____" ], [ "We can use `ls` to confirm that the file exists and check the size:", "_____no_output_____" ] ], [ [ "!ls -lh gd1_dataframe.hdf5", "-rw-rw-r-- 1 downey downey 17M Nov 18 19:06 gd1_dataframe.hdf5\r\n" ] ], [ [ "If you are using Windows, `ls` might not work; in that case, try:\n\n```\n!dir gd1_dataframe.hdf5\n```\n\nWe can read the file back like this:", "_____no_output_____" ] ], [ [ "read_back_df = pd.read_hdf(filename, 'df')\nread_back_df.shape", "_____no_output_____" ] ], [ [ "Pandas can write a variety of other formats, [which you can read about here](https://pandas.pydata.org/pandas-docs/stable/user_guide/io.html).", "_____no_output_____" ], [ "## Summary\n\nIn this lesson, we re-loaded the Gaia data we saved from a previous query.\n\nWe transformed the coordinates and proper motion from ICRS to a frame aligned with GD-1, and stored the results in a Pandas `DataFrame`.\n\nThen we replicated the selection process from the Price-Whelan and Bonaca paper:\n\n* We selected stars near the centerline of GD-1 and made a scatter plot of their proper motion.\n\n* We identified a region of proper motion that contains stars likely to be in GD-1.\n\n* We used a Boolean `Series` as a mask to select stars whose proper motion is in that region.\n\nSo far, we have used data from a relatively small region of the sky. In the next lesson, we'll write a query that selects stars based on proper motion, which will allow us to explore a larger region.", "_____no_output_____" ], [ "## Best practices\n\n* When you make a scatter plot, adjust the size of the markers and their transparency so the figure is not overplotted; otherwise it can misrepresent the data badly.\n\n* For simple scatter plots in Matplotlib, `plot` is faster than `scatter`.\n\n* An Astropy `Table` and a Pandas `DataFrame` are similar in many ways and they provide many of the same functions. They have pros and cons, but for many projects, either one would be a reasonable choice.", "_____no_output_____" ] ] ]

[ "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown" ]

[ [ "markdown", "markdown", "markdown" ], [ "code" ], [ "markdown" ], [ "code" ], [ "markdown" ], [ "code" ], [ "markdown" ], [ "code" ], [ "markdown" ], [ "code" ], [ "markdown" ], [ "code" ], [ "markdown" ], [ "code" ], [ "markdown" ], [ "code" ], [ "markdown" ], [ "code" ], [ "markdown" ], [ "code" ], [ "markdown" ], [ "code" ], [ "markdown", "markdown" ], [ "code" ], [ "markdown" ], [ "code" ], [ "markdown", "markdown", "markdown" ], [ "code" ], [ "markdown" ], [ "code" ], [ "markdown" ], [ "code" ], [ "markdown" ], [ "code" ], [ "markdown", "markdown", "markdown" ], [ "code" ], [ "markdown" ], [ "code" ], [ "markdown" ], [ "code" ], [ "markdown", "markdown" ], [ "code", "code", "code" ], [ "markdown", "markdown" ], [ "code" ], [ "markdown" ], [ "code" ], [ "markdown", "markdown" ], [ "code" ], [ "markdown" ], [ "code" ], [ "markdown", "markdown", "markdown" ], [ "code" ], [ "markdown" ], [ "code", "code" ], [ "markdown" ], [ "code" ], [ "markdown" ], [ "code" ], [ "markdown" ], [ "code" ], [ "markdown" ], [ "code" ], [ "markdown" ], [ "code" ], [ "markdown" ], [ "code" ], [ "markdown" ], [ "code" ], [ "markdown", "markdown" ], [ "code" ], [ "markdown" ], [ "code" ], [ "markdown" ], [ "code" ], [ "markdown" ], [ "code" ], [ "markdown" ], [ "code" ], [ "markdown" ], [ "code" ], [ "markdown" ], [ "code" ], [ "markdown" ], [ "code" ], [ "markdown", "markdown" ], [ "code" ], [ "markdown", "markdown" ], [ "code" ], [ "markdown", "markdown" ], [ "code" ], [ "markdown", "markdown" ], [ "code" ], [ "markdown" ], [ "code" ], [ "markdown", "markdown", "markdown" ] ]

[ "MIT" ]

[ "MIT" ]

[ "MIT" ]

[ [ [ "import warnings\n\nwarnings.filterwarnings('ignore')", "_____no_output_____" ], [ "import numpy as np\nimport pandas as pd\nimport matplotlib.pyplot as plt", "_____no_output_____" ], [ "import keras\nfrom keras.datasets import mnist\nfrom keras.layers import *\nfrom keras.layers.advanced_activations import LeakyReLU\nfrom keras.models import Sequential, Model\nfrom keras.optimizers import Adam", "_____no_output_____" ], [ "(X_train, _), (_, _) = mnist.load_data()", "_____no_output_____" ], [ "X_train[:5]", "_____no_output_____" ], [ "X_train.shape", "_____no_output_____" ], [ "X_train = (X_train-127.5)/127.5\n\nprint(X_train.min())\nprint(X_train.max())", "-1.0\n1.0\n" ], [ "TOTAL_EPOCHS = 50\nBATCH_SIZE = 256\nHALF_BATCH = 128\n\nNO_OF_BATCHES = int(X_train.shape[0]/BATCH_SIZE)\n\nNOISE_DIM = 100\n\nadam = Adam(lr = 2e-4, beta_1 = 0.5)", "_____no_output_____" ], [ "#Generator Modeling --> Up Sampling\n\ngenerator = Sequential()\ngenerator.add(Dense(units= 7*7*128, input_shape = (NOISE_DIM, )))\ngenerator.add(Reshape((7, 7, 128)))\ngenerator.add(LeakyReLU(0.2))\ngenerator.add(BatchNormalization())\n\n#(7, 7, 128) -> (14, 14, 64)\ngenerator.add(Conv2DTranspose(64, (3, 3), strides=(2, 2), padding='same'))\ngenerator.add(LeakyReLU(0.2))\ngenerator.add(BatchNormalization())\n\n#(14, 14, 64) --> (28, 28, 1)\ngenerator.add(Conv2DTranspose(1, (3, 3), strides=(2, 2), padding='same', activation='tanh'))\n\n\ngenerator.compile(loss = keras.losses.binary_crossentropy, optimizer=adam)\n\ngenerator.summary()", "Model: \"sequential\"\n_________________________________________________________________\nLayer (type) Output Shape Param # \n=================================================================\ndense (Dense) (None, 6272) 633472 \n_________________________________________________________________\nreshape (Reshape) (None, 7, 7, 128) 0 \n_________________________________________________________________\nleaky_re_lu (LeakyReLU) (None, 7, 7, 128) 0 \n_________________________________________________________________\nbatch_normalization (BatchNo (None, 7, 7, 128) 512 \n_________________________________________________________________\nconv2d_transpose (Conv2DTran (None, 14, 14, 64) 73792 \n_________________________________________________________________\nleaky_re_lu_1 (LeakyReLU) (None, 14, 14, 64) 0 \n_________________________________________________________________\nbatch_normalization_1 (Batch (None, 14, 14, 64) 256 \n_________________________________________________________________\nconv2d_transpose_1 (Conv2DTr (None, 28, 28, 1) 577 \n=================================================================\nTotal params: 708,609\nTrainable params: 708,225\nNon-trainable params: 384\n_________________________________________________________________\n" ], [ "#Discriminator Modeling --> Down Sampling\n\n# (28, 28, 1) --> (14, 14, 64)\ndiscriminator = Sequential()\ndiscriminator.add(Conv2D(64, kernel_size=(3, 3), strides=(2, 2), padding='same', input_shape = (28, 28, 1)))\ndiscriminator.add(LeakyReLU(0.2))\n\n#(14, 14, 64) --> (7, 7, 128)\ndiscriminator.add(Conv2D(128, kernel_size=(3, 3), strides=(2, 2), padding='same'))\ndiscriminator.add(LeakyReLU(0.2))\n\n#(7, 7, 128) --> 6272\ndiscriminator.add(Flatten())\ndiscriminator.add(Dense(100))\ndiscriminator.add(LeakyReLU(0.2))\n\ndiscriminator.add(Dense(1, activation='sigmoid'))\n\ndiscriminator.compile(loss = keras.losses.binary_crossentropy, optimizer=adam)\n\ndiscriminator.summary()", "Model: \"sequential_1\"\n_________________________________________________________________\nLayer (type) Output Shape Param # \n=================================================================\nconv2d (Conv2D) (None, 14, 14, 64) 640 \n_________________________________________________________________\nleaky_re_lu_2 (LeakyReLU) (None, 14, 14, 64) 0 \n_________________________________________________________________\nconv2d_1 (Conv2D) (None, 7, 7, 128) 73856 \n_________________________________________________________________\nleaky_re_lu_3 (LeakyReLU) (None, 7, 7, 128) 0 \n_________________________________________________________________\nflatten (Flatten) (None, 6272) 0 \n_________________________________________________________________\ndense_1 (Dense) (None, 100) 627300 \n_________________________________________________________________\nleaky_re_lu_4 (LeakyReLU) (None, 100) 0 \n_________________________________________________________________\ndense_2 (Dense) (None, 1) 101 \n=================================================================\nTotal params: 701,897\nTrainable params: 701,897\nNon-trainable params: 0\n_________________________________________________________________\n" ], [ "## Combied Model \n\ndiscriminator.trainable = False\ngan_input = Input(shape = (NOISE_DIM, ))\n\ngenerated_img = generator(gan_input)\n\ngan_output = discriminator(generated_img)\n\nmodel = Model(gan_input, gan_output)\n\nmodel.compile(loss = \"binary_crossentropy\", optimizer=adam)", "_____no_output_____" ], [ "model.summary()", "Model: \"model\"\n_________________________________________________________________\nLayer (type) Output Shape Param # \n=================================================================\ninput_1 (InputLayer) [(None, 100)] 0 \n_________________________________________________________________\nsequential (Sequential) (None, 28, 28, 1) 708609 \n_________________________________________________________________\nsequential_1 (Sequential) (None, 1) 701897 \n=================================================================\nTotal params: 1,410,506\nTrainable params: 708,225\nNon-trainable params: 702,281\n_________________________________________________________________\n" ], [ "#16.01\nX_train = X_train.reshape(-1, 28, 28, 1)", "_____no_output_____" ], [ "X_train.shape", "_____no_output_____" ], [ "def display_images(samples = 25):\n\n noise = np.random.normal(0, 1, size=(samples, NOISE_DIM))\n\n generated_img = generator.predict(noise)\n\n plt.figure(figsize=(10, 10))\n\n for i in range(samples):\n plt.subplot(5, 5, i+1)\n plt.imshow(generated_img[i].reshape(28, 28), cmap=\"gray\")\n plt.axis('off')\n\n plt.show()", "_____no_output_____" ], [ "## Training Loop\n\nd_losses = []\ng_losses = []\nfor epoch in range(TOTAL_EPOCHS):\n\n epoch_d_loss = 0.0\n epoch_g_loss = 0.0\n\n for step in range(NO_OF_BATCHES):\n\n\n #++++++++++++++++++#\n #Step1. :- Train Discriminator\n discriminator.trainable = True\n\n #get the real data\n idx = np.random.randint(0, 60000, HALF_BATCH)\n real_imgs = X_train[idx]\n\n #get the fake data\n noise = np.random.normal(0, 1, size=(HALF_BATCH, NOISE_DIM))\n fake_imgs = generator.predict(noise)\n\n #Labels\n real_y = np.ones((HALF_BATCH, 1))*0.9\n fake_y = np.zeros((HALF_BATCH, 1))\n\n #now train D\n d_loss_real = discriminator.train_on_batch(real_imgs, real_y)\n d_loss_fake = discriminator.train_on_batch(fake_imgs, fake_y)\n\n d_loss = 0.5*d_loss_real + 0.5*d_loss_fake\n\n epoch_d_loss += d_loss\n\n\n #++++++++++++++++++++++#\n #Step2. :- Train Generator\n discriminator.trainable = False\n\n noise = np.random.normal(0, 1, size=(BATCH_SIZE, NOISE_DIM))\n ground_truth_y = np.ones((BATCH_SIZE, 1))\n\n g_loss = model.train_on_batch(noise, ground_truth_y)\n\n epoch_g_loss += g_loss\n\n #++++++++++++++++++++++#\n\n print(\"Epoch : {}\\t, Discriminator Loss : {}\\t, Generator Loss : {}\".format((epoch+1), epoch_d_loss/NO_OF_BATCHES, epoch_g_loss/BATCH_SIZE))\n\n d_losses.append(epoch_d_loss/NO_OF_BATCHES)\n g_losses.append(epoch_g_loss/NO_OF_BATCHES)\n\n\n if (epoch+1)%10==0:\n generator.save('generator.h5')\n display_images()", "Epoch : 1\t, Discriminator Loss : 0.25897651991146986\t, Generator Loss : 1.1500072128983447\nEpoch : 2\t, Discriminator Loss : 0.5677873889923605\t, Generator Loss : 1.9773283661343157\nEpoch : 3\t, Discriminator Loss : 0.5564669755279509\t, Generator Loss : 1.508944047614932\nEpoch : 4\t, Discriminator Loss : 0.5703185591687504\t, Generator Loss : 1.4899448323994875\nEpoch : 5\t, Discriminator Loss : 0.621234845274534\t, Generator Loss : 1.3262150567024946\nEpoch : 6\t, Discriminator Loss : 0.6406578118475075\t, Generator Loss : 1.202875533606857\nEpoch : 7\t, Discriminator Loss : 0.642724542472607\t, Generator Loss : 1.1239283843897283\nEpoch : 8\t, Discriminator Loss : 0.6468475127958844\t, Generator Loss : 1.076513821259141\nEpoch : 9\t, Discriminator Loss : 0.6445728270416586\t, Generator Loss : 1.0509009454399347\nEpoch : 10\t, Discriminator Loss : 0.6408960496385893\t, Generator Loss : 1.0446941344998777\n" ], [ "", "_____no_output_____" ] ] ]

[ "code" ]

[ [ "code", "code", "code", "code", "code", "code", "code", "code", "code", "code", "code", "code", "code", "code", "code", "code", "code" ] ]

[ "MIT" ]

[ "MIT" ]

[ "MIT" ]

[ [ [ "# Preparation for Colab\n\nMake sure you're running a GPU runtime; if not, select \"GPU\" as the hardware accelerator in Runtime > Change Runtime Type in the menu. The next cells will install the `clip` package and its dependencies, and check if PyTorch 1.7.1 or later is installed.", "_____no_output_____" ] ], [ [ "# ! pip install ftfy regex tqdm\n# ! pip install git+https://github.com/openai/CLIP.git", "Collecting ftfy\n Downloading ftfy-6.0.3.tar.gz (64 kB)\n\u001b[?25l\r\u001b[K |█████ | 10 kB 14.9 MB/s eta 0:00:01\r\u001b[K |██████████▏ | 20 kB 18.7 MB/s eta 0:00:01\r\u001b[K |███████████████▎ | 30 kB 9.0 MB/s eta 0:00:01\r\u001b[K |████████████████████▍ | 40 kB 4.1 MB/s eta 0:00:01\r\u001b[K |█████████████████████████▌ | 51 kB 4.6 MB/s eta 0:00:01\r\u001b[K |██████████████████████████████▋ | 61 kB 4.7 MB/s eta 0:00:01\r\u001b[K |████████████████████████████████| 64 kB 1.3 MB/s \n\u001b[?25hRequirement already satisfied: regex in /usr/local/lib/python3.7/dist-packages (2019.12.20)\nRequirement already satisfied: tqdm in /usr/local/lib/python3.7/dist-packages (4.41.1)\nRequirement already satisfied: wcwidth in /usr/local/lib/python3.7/dist-packages (from ftfy) (0.2.5)\nBuilding wheels for collected packages: ftfy\n Building wheel for ftfy (setup.py) ... \u001b[?25l\u001b[?25hdone\n Created wheel for ftfy: filename=ftfy-6.0.3-py3-none-any.whl size=41934 sha256=90ec193331444b2c4ff1cd81935e7de42065b89d304db7efac67bcfd87c27873\n Stored in directory: /root/.cache/pip/wheels/19/f5/38/273eb3b5e76dfd850619312f693716ac4518b498f5ffb6f56d\nSuccessfully built ftfy\nInstalling collected packages: ftfy\nSuccessfully installed ftfy-6.0.3\nCollecting git+https://github.com/openai/CLIP.git\n Cloning https://github.com/openai/CLIP.git to /tmp/pip-req-build-hqnbveqi\n Running command git clone -q https://github.com/openai/CLIP.git /tmp/pip-req-build-hqnbveqi\nRequirement already satisfied: ftfy in /usr/local/lib/python3.7/dist-packages (from clip==1.0) (6.0.3)\nRequirement already satisfied: regex in /usr/local/lib/python3.7/dist-packages (from clip==1.0) (2019.12.20)\nRequirement already satisfied: tqdm in /usr/local/lib/python3.7/dist-packages (from clip==1.0) (4.41.1)\nRequirement already satisfied: torch in /usr/local/lib/python3.7/dist-packages (from clip==1.0) (1.9.0+cu102)\nRequirement already satisfied: torchvision in /usr/local/lib/python3.7/dist-packages (from clip==1.0) (0.10.0+cu102)\nRequirement already satisfied: wcwidth in /usr/local/lib/python3.7/dist-packages (from ftfy->clip==1.0) (0.2.5)\nRequirement already satisfied: typing-extensions in /usr/local/lib/python3.7/dist-packages (from torch->clip==1.0) (3.7.4.3)\nRequirement already satisfied: numpy in /usr/local/lib/python3.7/dist-packages (from torchvision->clip==1.0) (1.19.5)\nRequirement already satisfied: pillow>=5.3.0 in /usr/local/lib/python3.7/dist-packages (from torchvision->clip==1.0) (7.1.2)\nBuilding wheels for collected packages: clip\n Building wheel for clip (setup.py) ... \u001b[?25l\u001b[?25hdone\n Created wheel for clip: filename=clip-1.0-py3-none-any.whl size=1369080 sha256=fda43d2b80cfb2b33c2d43e23ea5f53293a9a8b48d5f9e341de527f6adfbf5a3\n Stored in directory: /tmp/pip-ephem-wheel-cache-kmmplf44/wheels/fd/b9/c3/5b4470e35ed76e174bff77c92f91da82098d5e35fd5bc8cdac\nSuccessfully built clip\nInstalling collected packages: clip\nSuccessfully installed clip-1.0\n" ], [ "import numpy as np\nimport torch\nimport clip\n# from tqdm.notebook import tqdm\nfrom tqdm import tqdm\n\nprint(\"Torch version:\", torch.__version__)\n\nassert torch.__version__.split(\".\") >= [\"1\", \"7\", \"1\"], \"PyTorch 1.7.1 or later is required\"", "Torch version: 1.9.0\n" ] ], [ [ "# Loading the model\n\nDownload and instantiate a CLIP model using the `clip` module that we just installed.", "_____no_output_____" ] ], [ [ "print(clip.available_models())\nprint(os.getcwd())", "['RN50', 'RN101', 'RN50x4', 'RN50x16', 'ViT-B/32', 'ViT-B/16']\n/localdata/workspace/CLIP/notebooks\n" ], [ "# model, preprocess = clip.load(\"ViT-B/32\")\nmodel, preprocess = clip.load(\"../models/200m0.988.pt\")", "_____no_output_____" ], [ "input_resolution = model.visual.input_resolution\ncontext_length = model.context_length\nvocab_size = model.vocab_size\n\nprint(\"Model parameters:\", f\"{np.sum([int(np.prod(p.shape)) for p in model.parameters()]):,}\")\nprint(\"Input resolution:\", input_resolution)\nprint(\"Context length:\", context_length)\nprint(\"Vocab size:\", vocab_size)", "Model parameters: 151,277,313\nInput resolution: 224\nContext length: 77\nVocab size: 49408\n" ] ], [ [ "# Preparing ImageNet labels and prompts\n\nThe following cell contains the 1,000 labels for the ImageNet dataset, followed by the text templates we'll use as \"prompt engineering\".", "_____no_output_____" ] ], [ [ "classes = [\n 'apple',\n 'aquarium fish',\n 'baby',\n 'bear',\n 'beaver',\n 'bed',\n 'bee',\n 'beetle',\n 'bicycle',\n 'bottle',\n 'bowl',\n 'boy',\n 'bridge',\n 'bus',\n 'butterfly',\n 'camel',\n 'can',\n 'castle',\n 'caterpillar',\n 'cattle',\n 'chair',\n 'chimpanzee',\n 'clock',\n 'cloud',\n 'cockroach',\n 'couch',\n 'crab',\n 'crocodile',\n 'cup',\n 'dinosaur',\n 'dolphin',\n 'elephant',\n 'flatfish',\n 'forest',\n 'fox',\n 'girl',\n 'hamster',\n 'house',\n 'kangaroo',\n 'keyboard',\n 'lamp',\n 'lawn mower',\n 'leopard',\n 'lion',\n 'lizard',\n 'lobster',\n 'man',\n 'maple tree',\n 'motorcycle',\n 'mountain',\n 'mouse',\n 'mushroom',\n 'oak tree',\n 'orange',\n 'orchid',\n 'otter',\n 'palm tree',\n 'pear',\n 'pickup truck',\n 'pine tree',\n 'plain',\n 'plate',\n 'poppy',\n 'porcupine',\n 'possum',\n 'rabbit',\n 'raccoon',\n 'ray',\n 'road',\n 'rocket',\n 'rose',\n 'sea',\n 'seal',\n 'shark',\n 'shrew',\n 'skunk',\n 'skyscraper',\n 'snail',\n 'snake',\n 'spider',\n 'squirrel',\n 'streetcar',\n 'sunflower',\n 'sweet pepper',\n 'table',\n 'tank',\n 'telephone',\n 'television',\n 'tiger',\n 'tractor',\n 'train',\n 'trout',\n 'tulip',\n 'turtle',\n 'wardrobe',\n 'whale',\n 'willow tree',\n 'wolf',\n 'woman',\n 'worm',\n]\n\ntemplates = [\n 'a photo of a {}.',\n 'a blurry photo of a {}.',\n 'a black and white photo of a {}.',\n 'a low contrast photo of a {}.',\n 'a high contrast photo of a {}.',\n 'a bad photo of a {}.',\n 'a good photo of a {}.',\n 'a photo of a small {}.',\n 'a photo of a big {}.',\n 'a photo of the {}.',\n 'a blurry photo of the {}.',\n 'a black and white photo of the {}.',\n 'a low contrast photo of the {}.',\n 'a high contrast photo of the {}.',\n 'a bad photo of the {}.',\n 'a good photo of the {}.',\n 'a photo of the small {}.',\n 'a photo of the big {}.',\n]\n", "_____no_output_____" ] ], [ [ "A subset of these class names are modified from the default ImageNet class names sourced from Anish Athalye's imagenet-simple-labels.\n\nThese edits were made via trial and error and concentrated on the lowest performing classes according to top_1 and top_5 accuracy on the ImageNet training set for the RN50, RN101, and RN50x4 models. These tweaks improve top_1 by 1.5% on ViT-B/32 over using the default class names. Alec got bored somewhere along the way as gains started to diminish and never finished updating / tweaking the list. He also didn't revisit this with the better performing RN50x16, RN50x64, or any of the ViT models. He thinks it's likely another 0.5% to 1% top_1 could be gained from further work here. It'd be interesting to more rigorously study / understand this.\n\nSome examples beyond the crane/crane -> construction crane / bird crane issue mentioned in Section 3.1.4 of the paper include:\n\n- CLIP interprets \"nail\" as \"fingernail\" so we changed the label to \"metal nail\".\n- ImageNet kite class refers to the bird of prey, not the flying toy, so we changed \"kite\" to \"kite (bird of prey)\"\n- The ImageNet class for red wolf seems to include a lot of mislabeled maned wolfs so we changed \"red wolf\" to \"red wolf or maned wolf\"", "_____no_output_____" ] ], [ [ "\n\nprint(f\"{len(classes)} classes, {len(templates)} templates\")", "100 classes, 18 templates\n" ] ], [ [ "A similar, intuition-guided trial and error based on the ImageNet training set was used for templates. This list is pretty haphazard and was gradually made / expanded over the course of about a year of the project and was revisited / tweaked every few months. A surprising / weird thing was adding templates intended to help ImageNet-R performance (specifying different possible renditions of an object) improved standard ImageNet accuracy too.\n\nAfter the 80 templates were \"locked\" for the paper, we ran sequential forward selection over the list of 80 templates. The search terminated after ensembling 7 templates and selected them in the order below.\n\n1. itap of a {}.\n2. a bad photo of the {}.\n3. a origami {}.\n4. a photo of the large {}.\n5. a {} in a video game.\n6. art of the {}.\n7. a photo of the small {}.\n\nSpeculating, we think it's interesting to see different scales (large and small), a difficult view (a bad photo), and \"abstract\" versions (origami, video game, art), were all selected for, but we haven't studied this in any detail. This subset performs a bit better than the full 80 ensemble reported in the paper, especially for the smaller models.", "_____no_output_____" ], [ "# Loading the Images\n\nThe ILSVRC2012 datasets are no longer available for download publicly. We instead download the ImageNet-V2 dataset by [Recht et al.](https://arxiv.org/abs/1902.10811).\n\nIf you have the ImageNet dataset downloaded, you can replace the dataset with the official torchvision loader, e.g.:\n\n```python\nimages = torchvision.datasets.ImageNet(\"path/to/imagenet\", split='val', transform=preprocess)\n```", "_____no_output_____" ] ], [ [ "# ! pip install git+https://github.com/modestyachts/ImageNetV2_pytorch\n\n# from imagenetv2_pytorch import ImageNetV2Dataset\n\n# images = ImageNetV2Dataset(transform=preprocess)\n\nfrom torchvision.datasets import CIFAR100\nimport os\ncifar100 = CIFAR100(root=os.path.expanduser(\"./data/cifar100\"), download=True, train=True, transform=preprocess)\n\nloader = torch.utils.data.DataLoader(cifar100, batch_size=32, num_workers=8)", "Files already downloaded and verified\n" ] ], [ [ "# Creating zero-shot classifier weights", "_____no_output_____" ] ], [ [ "def zeroshot_classifier(classnames, templates):\n with torch.no_grad():\n zeroshot_weights = []\n for classname in tqdm(classnames):\n texts = [template.format(classname) for template in templates] #format with class\n texts = clip.tokenize(texts).cuda() #tokenize\n class_embeddings = model.encode_text(texts) #embed with text encoder\n class_embeddings /= class_embeddings.norm(dim=-1, keepdim=True)\n class_embedding = class_embeddings.mean(dim=0)\n class_embedding /= class_embedding.norm()\n zeroshot_weights.append(class_embedding)\n zeroshot_weights = torch.stack(zeroshot_weights, dim=1).cuda()\n return zeroshot_weights\n\n\nzeroshot_weights = zeroshot_classifier(classes, templates)", "100%|██████████| 100/100 [00:01<00:00, 74.94it/s]\n" ] ], [ [ "# Zero-shot prediction", "_____no_output_____" ] ], [ [ "def accuracy(output, target, topk=(1,)):\n pred = output.topk(max(topk), 1, True, True)[1].t()\n correct = pred.eq(target.view(1, -1).expand_as(pred))\n return [float(correct[:k].reshape(-1).float().sum(0, keepdim=True).cpu().numpy()) for k in topk]", "_____no_output_____" ], [ "with torch.no_grad():\n top1, top5, n = 0., 0., 0.\n for i, (images, target) in enumerate(tqdm(loader)):\n images = images.cuda()\n target = target.cuda()\n \n # predict\n image_features = model.encode_image(images)\n image_features /= image_features.norm(dim=-1, keepdim=True)\n logits = 100. * image_features @ zeroshot_weights\n\n # measure accuracy\n acc1, acc5 = accuracy(logits, target, topk=(1, 5))\n top1 += acc1\n top5 += acc5\n n += images.size(0)\n\ntop1 = (top1 / n) * 100\ntop5 = (top5 / n) * 100 \n\nprint(f\"Top-1 accuracy: {top1:.2f}\")\nprint(f\"Top-5 accuracy: {top5:.2f}\")", "100%|██████████| 1563/1563 [00:43<00:00, 35.92it/s]" ] ] ]

[ "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code" ]

[ [ "markdown" ], [ "code", "code" ], [ "markdown" ], [ "code", "code", "code" ], [ "markdown" ], [ "code" ], [ "markdown" ], [ "code" ], [ "markdown", "markdown" ], [ "code" ], [ "markdown" ], [ "code" ], [ "markdown" ], [ "code", "code" ] ]

[ "AAL" ]

[ "AAL" ]

[ "AAL" ]

[ [ [ "## Car Accidents During COVID Stay-At-Home Periods Analysis\nThis notebook is using data collected and cleaned in the `car_accidents_eda.ipynb` notebook, and will be used for examining differences state-by-state for severity and frequency of car accidents. Tableau graphs will be used for final graphs, but this notebook gives a sense of direction.", "_____no_output_____" ], [ "### TABLE OF CONTENTS\n1. [Setup](#Setup)\n2. [Top 20 States Analysis](#Top-20-States-Analysis)\n - [California](#California)\n - [Texas](#Texas)\n - [Florida](#Florida)\n - [South Carolina](#South-Carolina)\n - [North Carolina](#North-Carolina)\n - [New York](#New-York)\n - [Pennsylvania](#Pennsylvania)\n - [Illinois](#Illinois)\n - [Virginia](#Virginia)\n - [Michigan](#Michigan)\n - [Georgia](#Georgia)\n - [Oregon](#Oregon)\n - [Minnesota](#Minnesota)\n - [Arizona](#Arizona)\n - [Tennessee](#Tennessee)\n - [Washington](#Washington)\n - [Ohio](#Ohio)\n - [Louisiana](#Louisiana)\n - [Oklahoma](#Oklahoma)\n - [New Jersey](#New-Jersey)\n3. [Top 20 States Averages](#Top-20-States-Averages)", "_____no_output_____" ], [ "First, import libraries and modules needed >", "_____no_output_____" ] ], [ [ "import pickle\nimport pandas as pd\nimport numpy as np", "_____no_output_____" ] ], [ [ "### Setup", "_____no_output_____" ], [ "First import the dataset created and prepared for analysis >", "_____no_output_____" ] ], [ [ "# open our master dataset for analysis\n\nwith open('pickle/car_accidents_master.pickle','rb') as read_file:\n car_accidents_master = pickle.load(read_file)", "_____no_output_____" ] ], [ [ "Note that we are looking at our dataset before it was narrowed for modeling (e.g. only severity 2 & 3, only using mapquest-sourced data). For the analysis we'll be doing here we can look at the full dataset.\n\nBut first we'll narrow to only the columns we'll be using, which are date, state, whether in period of shutdown, and severity of the accident >", "_____no_output_____" ] ], [ [ "accident_covid = car_accidents_master[['Severity','Date',\n 'State','Year','Month',\n 'Day','Day_Of_Week',\n 'Shut_Down'\n ]]", "_____no_output_____" ] ], [ [ "Now we'll look at amount of accidents in each states >", "_____no_output_____" ] ], [ [ "accident_covid.State.value_counts()", "_____no_output_____" ] ], [ [ "And then see if we narrowed the number of states, how many we'd need to get to at least 75% of total accidents >", "_____no_output_____" ] ], [ [ "accident_covid.State.value_counts()[:20].sum()/accident_covid.State.value_counts().sum()", "_____no_output_____" ], [ "accident_covid.State.value_counts()[:20]", "_____no_output_____" ] ], [ [ "First we'll focus on the state with the highest amount of accidents, as they will provided more data to then draw better conclusions from and also are of greater interest since more need to address car accidents. We'll focus on the 20 states that have the most car accidents, which also represent 86% of total accidents in our full dataset for the US. Those states are: CA, TX, FL, SC, NC, NY, PA, IL, VA, MI, GA, OR, MN, AZ, TN, WA, OH, LA, OK, and NJ. These states also vary in time period of shutdown and political leanings that may influence behavior around a shutdown period, so we have nice diversity for comparison.\n\nFor each state, we'll create dataframes for separate periods we want to analyze and compare. We'll compare stay-at-home order periods with past three years in that same period.\n\nFirst we'll create functions to get the data we'll need for each state >", "_____no_output_____" ] ], [ [ "def severity_percent(state_data):\n '''\n Takes in state data and returns\n a list of percentages for each \n severity level to total accidents.\n '''\n percentages = []\n if 1 not in sorted(state_data.Severity.unique()):\n percentages.append(0)\n \n for x in sorted(state_data.Severity.unique()):\n percent = state_data.Severity.value_counts()[x]/state_data.Severity.count()\n percentages.append(round((percent*100),2))\n \n if 4 not in sorted(state_data.Severity.unique()):\n percentages.append(0)\n \n return percentages", "_____no_output_____" ], [ "def change_freq(state_freq_data):\n '''\n Takes in the dataframe for fequency of \n car accidents for a state and outputs\n the percent change from stay-home 2020\n period compared to average of equivalent\n periods 2017-2019 rounded to 2 decimals.\n '''\n freq_2020 = state_freq_data.Num_Accidents[1]\n avg_freq = np.mean(state_freq_data.Num_Accidents[2:5])\n \n return round(((freq_2020 - avg_freq)/avg_freq * 100),2)", "_____no_output_____" ] ], [ [ "-\n## Top 20 States Analysis", "_____no_output_____" ], [ "### California", "_____no_output_____" ], [ "First create periods we'll want to compare for the state >", "_____no_output_____" ] ], [ [ "# full state period\nca_ttl = accident_covid[(accident_covid.State == 'CA')]\n\n# shutdown period\nca_sd = accident_covid[\n (accident_covid.State == 'CA') & \n (accident_covid.Shut_Down == 1)]\n\n# 2019 \n# note CA is still in shutdown, so we're comparing to when dataset ended in 2020\n# not when their stay-at-home orders ended\n\nca_2019 = accident_covid[(car_accidents_master['Date'] > '2019-03-19') & \n (car_accidents_master['Date'] <= '2019-06-30') & \n (car_accidents_master['State'] == 'CA')]\n\n# 2018\nca_2018 = accident_covid[(car_accidents_master['Date'] > '2018-03-19') & \n (car_accidents_master['Date'] <= '2018-06-30') & \n (car_accidents_master['State'] == 'CA')]\n\n# 2017\nca_2017 = accident_covid[(car_accidents_master['Date'] > '2017-03-19') & \n (car_accidents_master['Date'] <= '2017-06-30') & \n (car_accidents_master['State'] == 'CA')]", "_____no_output_____" ] ], [ [ "#### Frequency\nNow compare frequencies (total number) of car accidents for total, shut-down period in 2020, 2019, 2018, and 2017 (2019, 2018, & 2017 being in same month-day date range as shut-down period) >", "_____no_output_____" ] ], [ [ "# find total frequencies for state total and in each period \n\nca_freqs = [ca_ttl.shape[0],\n ca_sd.shape[0],\n ca_2019.shape[0],\n ca_2018.shape[0],\n ca_2017.shape[0]\n ]\n\n# then develop a simple dataframe to compare them\nca_freq_data = pd.DataFrame({'Timeframe': ['Total','Shut_Down',2019,2018,2017],\n 'Num_Accidents': ca_freqs})\n\nca_freq_data", "_____no_output_____" ], [ "# percent change in frequency in 2020 stay-home compared to average past years\n\nchange = change_freq(ca_freq_data)\nprint(\"Change Frequency: %.2f%%\" % (change))", "Change Frequency: 76.44%\n" ] ], [ [ "#### Severity\nNow compare severities of car accidents for total, shut-down period in 2020, 2019, 2018, and 2017 (2019, 2018 & 2017 being in same month-day date range as shut-down period) >", "_____no_output_____" ] ], [ [ "# first find percentages of severity ranks to total accidents for \n# total and for each period \n\nca_sp_ttl = severity_percent(ca_ttl)\nca_sp_sd = severity_percent(ca_sd)\nca_sp_2019 = severity_percent(ca_2019)\nca_sp_2018 = severity_percent(ca_2018)\nca_sp_2017 = severity_percent(ca_2017)\n\n# then again create a simple dataframe for comparison\n\nca_severity_data = pd.DataFrame({'Severity_Rank': [1,2,3,4],\n 'TTL': ca_sp_ttl,\n 'Shut_Down': ca_sp_sd,\n '2019': ca_sp_2019,\n '2018': ca_sp_2018,\n '2017': ca_sp_2017,\n })\n\nca_severity_data", "_____no_output_____" ] ], [ [ "#### CA Takeaways:\n- See increase in fequency during shut down period compared to same periods in past three years\n- Seeing increase in 'edges' of severity -- 1 & 4 -- of car accidents with lowering of 'middle' severities -- 2 & 3 -- for shutdown period compared to average ttl and past year's prior periods", "_____no_output_____" ], [ "-\n### Texas", "_____no_output_____" ], [ "First create periods we'll want to compare for the state >", "_____no_output_____" ] ], [ [ "# full state period\ntx_ttl = accident_covid[(accident_covid.State == 'TX')]\n\n# shutdown period\ntx_sd = accident_covid[\n (accident_covid.State == 'TX') & \n (accident_covid.Shut_Down == 1)]\n\n# 2019 \ntx_2019 = accident_covid[(car_accidents_master['Date'] > '2019-03-31') & \n (car_accidents_master['Date'] <= '2019-04-30') & \n (car_accidents_master['State'] == 'TX')]\n\n# 2018\ntx_2018 = accident_covid[(car_accidents_master['Date'] > '2018-03-31') & \n (car_accidents_master['Date'] <= '2018-04-30') & \n (car_accidents_master['State'] == 'TX')]\n\n# 2017\ntx_2017 = accident_covid[(car_accidents_master['Date'] > '2017-03-31') & \n (car_accidents_master['Date'] <= '2017-04-30') & \n (car_accidents_master['State'] == 'TX')]", "_____no_output_____" ] ], [ [ "#### Frequency\nNow compare frequencies (total number) of car accidents for total, shut-down period in 2020, 2019, 2018, and 2017 (2019, 2018, & 2017 being in same month-day date range as shut-down period) >", "_____no_output_____" ] ], [ [ "# find total frequencies for state total and in each period \n\ntx_freqs = [tx_ttl.shape[0],\n tx_sd.shape[0],\n tx_2019.shape[0],\n tx_2018.shape[0],\n tx_2017.shape[0]\n ]\n\n# then develop a simple dataframe to compare them\ntx_freq_data = pd.DataFrame({'Timeframe': ['Total','Shut_Down',2019,2018,2017],\n 'Num_Accidents': tx_freqs})\n\ntx_freq_data", "_____no_output_____" ], [ "# percent change in frequency in 2020 stay-home compared to average past years\n\nchange = change_freq(tx_freq_data)\nprint(\"Change Frequency: %.2f%%\" % (change))", "Change Frequency: -37.65%\n" ] ], [ [ "#### Severity\nNow compare severities of car accidents for total, shut-down period in 2020, 2019, 2018, and 2017 (2019, 2018 & 2017 being in same month-day date range as shut-down period) >", "_____no_output_____" ] ], [ [ "# first find percentages of severity ranks to total accidents for \n# total and for each period \n\ntx_sp_ttl = severity_percent(tx_ttl)\ntx_sp_sd = severity_percent(tx_sd)\ntx_sp_2019 = severity_percent(tx_2019)\ntx_sp_2018 = severity_percent(tx_2018)\ntx_sp_2017 = severity_percent(tx_2017)\n\n# then again create a simple dataframe for comparison\n\ntx_severity_data = pd.DataFrame({'Severity_Rank': [1,2,3,4],\n 'TTL': tx_sp_ttl,\n 'Shut_Down': tx_sp_sd,\n '2019': tx_sp_2019,\n '2018': tx_sp_2018,\n '2017': tx_sp_2017\n })\n\ntx_severity_data", "_____no_output_____" ] ], [ [ "#### TX Takeaways:\n- Frequency: Unlike CA, TX had a drop in frequency during shutdown compared to past periods\n- Severity: Similar to CA, still the 'edges' of severity (1 & 4) growing, 4 in particular to a greater degree than CA ", "_____no_output_____" ], [ "-\n### Florida\n\nFirst create periods we'll want to compare for the state >", "_____no_output_____" ] ], [ [ "# full state period\nfl_ttl = accident_covid[(accident_covid.State == 'FL')]\n\n# shutdown period\nfl_sd = accident_covid[\n (accident_covid.State == 'FL') & \n (accident_covid.Shut_Down == 1)]\n\n# 2019 \nfl_2019 = accident_covid[(car_accidents_master['Date'] > '2019-04-03') & \n (car_accidents_master['Date'] <= '2019-04-30') & \n (car_accidents_master['State'] == 'FL')]\n\n# 2018\nfl_2018 = accident_covid[(car_accidents_master['Date'] > '2018-04-03') & \n (car_accidents_master['Date'] <= '2018-04-30') & \n (car_accidents_master['State'] == 'FL')]\n\n# 2017\nfl_2017 = accident_covid[(car_accidents_master['Date'] > '2017-04-03') & \n (car_accidents_master['Date'] <= '2017-04-30') & \n (car_accidents_master['State'] == 'FL')]", "_____no_output_____" ] ], [ [ "#### Frequency\nNow compare frequencies (total number) of car accidents for total, shut-down period in 2020, 2019, 2018, and 2017 (2019, 2018, & 2017 being in same month-day date range as shut-down period) >", "_____no_output_____" ] ], [ [ "# find total frequencies for state total and in each period \n\nfl_freqs = [fl_ttl.shape[0],\n fl_sd.shape[0],\n fl_2019.shape[0],\n fl_2018.shape[0],\n fl_2017.shape[0]\n ]\n\n# then develop a simple dataframe to compare them\nfl_freq_data = pd.DataFrame({'Timeframe': ['Total','Shut_Down',2019,2018,2017],\n 'Num_Accidents': fl_freqs})\n\nfl_freq_data", "_____no_output_____" ], [ "# percent change in frequency in 2020 stay-home compared to average past years\n\nchange = change_freq(fl_freq_data)\nprint(\"Change Frequency: %.2f%%\" % (change))", "Change Frequency: 6.62%\n" ] ], [ [ "#### Severity\nNow compare severities of car accidents for total, shut-down period in 2020, 2019, 2018, and 2017 (2019, 2018 & 2017 being in same month-day date range as shut-down period) >", "_____no_output_____" ] ], [ [ "# first find percentages of severity ranks to total accidents for \n# total and for each period \n\nfl_sp_ttl = severity_percent(fl_ttl)\nfl_sp_sd = severity_percent(fl_sd)\nfl_sp_2019 = severity_percent(fl_2019)\nfl_sp_2018 = severity_percent(fl_2018)\nfl_sp_2017 = severity_percent(fl_2017)\n\n# then again create a simple dataframe for comparison\n\nfl_severity_data = pd.DataFrame({'Severity_Rank': [1,2,3,4],\n 'TTL': fl_sp_ttl,\n 'Shut_Down': fl_sp_sd,\n '2019': fl_sp_2019,\n '2018': fl_sp_2018,\n '2017': fl_sp_2017\n })\n\nfl_severity_data", "_____no_output_____" ] ], [ [ "#### FL Takeaways:\n- Frequency: Not really much change positively or negatively for shutdown compared to past periods\n- Severity: Don't see a lot of difference in severity 4 as with CA & TX, but do see significant increase in 1s while 2 & 3 dropped", "_____no_output_____" ], [ "-\n### South Carolina\n\nFirst create periods we'll want to compare for the state >", "_____no_output_____" ] ], [ [ "# full state period\nsc_ttl = accident_covid[(accident_covid.State == 'SC')]\n\n# shutdown period\nsc_sd = accident_covid[\n (accident_covid.State == 'SC') & \n (accident_covid.Shut_Down == 1)]\n\n# 2019 \nsc_2019 = accident_covid[(car_accidents_master['Date'] > '2019-04-06') & \n (car_accidents_master['Date'] <= '2019-05-04') & \n (car_accidents_master['State'] == 'SC')]\n\n# 2018\nsc_2018 = accident_covid[(car_accidents_master['Date'] > '2018-04-06') & \n (car_accidents_master['Date'] <= '2018-05-04') & \n (car_accidents_master['State'] == 'SC')]\n\n# 2017\nsc_2017 = accident_covid[(car_accidents_master['Date'] > '2017-04-06') & \n (car_accidents_master['Date'] <= '2017-05-04') & \n (car_accidents_master['State'] == 'SC')]", "_____no_output_____" ] ], [ [ "#### Frequency\nNow compare frequencies (total number) of car accidents for total, shut-down period in 2020, 2019, 2018, and 2017 (2019, 2018, & 2017 being in same month-day date range as shut-down period) >", "_____no_output_____" ] ], [ [ "# find total frequencies for state total and in each period \n\nsc_freqs = [sc_ttl.shape[0],\n sc_sd.shape[0],\n sc_2019.shape[0],\n sc_2018.shape[0],\n sc_2017.shape[0]\n ]\n\n# then develop a simple dataframe to compare them\nsc_freq_data = pd.DataFrame({'Timeframe': ['Total','Shut_Down',2019,2018,2017],\n 'Num_Accidents': sc_freqs})\n\nsc_freq_data", "_____no_output_____" ], [ "# percent change in frequency in 2020 stay-home compared to average past years\n\nchange = change_freq(sc_freq_data)\nprint(\"Change Frequency: %.2f%%\" % (change))", "Change Frequency: 23.58%\n" ] ], [ [ "#### Severity\nNow compare severities of car accidents for total, shut-down period in 2020, 2019, 2018, and 2017 (2019, 2018 & 2017 being in same month-day date range as shut-down period) >", "_____no_output_____" ] ], [ [ "# first find percentages of severity ranks to total accidents for \n# total and for each period \n\nsc_sp_ttl = severity_percent(sc_ttl)\nsc_sp_sd = severity_percent(sc_sd)\nsc_sp_2019 = severity_percent(sc_2019)\nsc_sp_2018 = severity_percent(sc_2018)\nsc_sp_2017 = severity_percent(sc_2017)\n\n# then again create a simple dataframe for comparison\n\nsc_severity_data = pd.DataFrame({'Severity_Rank': [1,2,3,4],\n 'TTL': sc_sp_ttl,\n 'Shut_Down': sc_sp_sd,\n '2019': sc_sp_2019,\n '2018': sc_sp_2018,\n '2017': sc_sp_2017\n })\n\nsc_severity_data", "_____no_output_____" ] ], [ [ "#### SC Takeaways:\n- Frequency: no significant difference for shutdown period. 2017 seems low, but this may be due to less data collected that year. \n- Severity: in general, seemed to lower in severity. See more 1 & 2 and less 3 & 4", "_____no_output_____" ], [ "-\n### North Carolina\n\nFirst create periods we'll want to compare for the state >", "_____no_output_____" ] ], [ [ "# full state period\nnc_ttl = accident_covid[(accident_covid.State == 'NC')]\n\n# shutdown period\nnc_sd = accident_covid[\n (accident_covid.State == 'NC') & \n (accident_covid.Shut_Down == 1)]\n\n# 2019 \nnc_2019 = accident_covid[(car_accidents_master['Date'] > '2019-03-30') & \n (car_accidents_master['Date'] <= '2019-05-22') & \n (car_accidents_master['State'] == 'NC')]\n\n# 2018\nnc_2018 = accident_covid[(car_accidents_master['Date'] > '2018-03-30') & \n (car_accidents_master['Date'] <= '2018-05-22') & \n (car_accidents_master['State'] == 'NC')]\n\n# 2017\nnc_2017 = accident_covid[(car_accidents_master['Date'] > '2017-03-30') & \n (car_accidents_master['Date'] <= '2017-05-22') & \n (car_accidents_master['State'] == 'NC')]", "_____no_output_____" ] ], [ [ "#### Frequency\nNow compare frequencies (total number) of car accidents for total, shut-down period in 2020, 2019, 2018, and 2017 (2019, 2018, & 2017 being in same month-day date range as shut-down period) >", "_____no_output_____" ] ], [ [ "# find total frequencies for state total and in each period \n\nnc_freqs = [nc_ttl.shape[0],\n nc_sd.shape[0],\n nc_2019.shape[0],\n nc_2018.shape[0],\n nc_2017.shape[0]\n ]\n\n# then develop a simple dataframe to compare them\nnc_freq_data = pd.DataFrame({'Timeframe': ['Total','Shut_Down',2019,2018,2017],\n 'Num_Accidents': nc_freqs})\n\nnc_freq_data", "_____no_output_____" ], [ "# percent change in frequency in 2020 stay-home compared to average past years\n\nchange = change_freq(nc_freq_data)\nprint(\"Change Frequency: %.2f%%\" % (change))", "Change Frequency: 22.40%\n" ] ], [ [ "#### Severity\nNow compare severities of car accidents for total, shut-down period in 2020, 2019, 2018, and 2017 (2019, 2018 & 2017 being in same month-day date range as shut-down period) >", "_____no_output_____" ] ], [ [ "# first find percentages of severity ranks to total accidents for \n# total and for each period \n\nnc_sp_ttl = severity_percent(nc_ttl)\nnc_sp_sd = severity_percent(nc_sd)\nnc_sp_2019 = severity_percent(nc_2019)\nnc_sp_2018 = severity_percent(nc_2018)\nnc_sp_2017 = severity_percent(nc_2017)\n\n# then again create a simple dataframe for comparison\n\nnc_severity_data = pd.DataFrame({'Severity_Rank': [1,2,3,4],\n 'TTL': nc_sp_ttl,\n 'Shut_Down': nc_sp_sd,\n '2019': nc_sp_2019,\n '2018': nc_sp_2018,\n '2017': nc_sp_2017\n })\n\nnc_severity_data", "_____no_output_____" ] ], [ [ "#### NC Takeaways:\n- Fequency: No significant change during shutdown period\n- Severity: Do see an increase in level 4 severity, but greatest change is increase in severity 1 with drop in severity 2", "_____no_output_____" ], [ "-\n### New York\n\nFirst create periods we'll want to compare for the state >", "_____no_output_____" ] ], [ [ "# full state period\nny_ttl = accident_covid[(accident_covid.State == 'NY')]\n\n# shutdown period\nny_sd = accident_covid[\n (accident_covid.State == 'NY') & \n (accident_covid.Shut_Down == 1)]\n\n# 2019 \nny_2019 = accident_covid[(car_accidents_master['Date'] > '2019-03-22') & \n (car_accidents_master['Date'] <= '2019-06-13') & \n (car_accidents_master['State'] == 'NY')]\n\n# 2018\nny_2018 = accident_covid[(car_accidents_master['Date'] > '2018-03-22') & \n (car_accidents_master['Date'] <= '2018-06-13') & \n (car_accidents_master['State'] == 'NY')]\n\n# 2017\nny_2017 = accident_covid[(car_accidents_master['Date'] > '2017-03-22') & \n (car_accidents_master['Date'] <= '2017-06-13') & \n (car_accidents_master['State'] == 'NY')]", "_____no_output_____" ] ], [ [ "#### Frequency\nNow compare frequencies (total number) of car accidents for total, shut-down period in 2020, 2019, 2018, and 2017 (2019, 2018, & 2017 being in same month-day date range as shut-down period) >", "_____no_output_____" ] ], [ [ "# find total frequencies for state total and in each period \n\nny_freqs = [ny_ttl.shape[0],\n ny_sd.shape[0],\n ny_2019.shape[0],\n ny_2018.shape[0],\n ny_2017.shape[0]\n ]\n\n# then develop a simple dataframe to compare them\nny_freq_data = pd.DataFrame({'Timeframe': ['Total','Shut_Down',2019,2018,2017],\n 'Num_Accidents': ny_freqs})\n\nny_freq_data", "_____no_output_____" ], [ "# percent change in frequency in 2020 stay-home compared to average past years\n\nchange = change_freq(ny_freq_data)\nprint(\"Change Frequency: %.2f%%\" % (change))", "Change Frequency: 50.28%\n" ] ], [ [ "#### Severity\nNow compare severities of car accidents for total, shut-down period in 2020, 2019, 2018, and 2017 (2019, 2018 & 2017 being in same month-day date range as shut-down period) >", "_____no_output_____" ] ], [ [ "# first find percentages of severity ranks to total accidents for \n# total and for each period \n\nny_sp_ttl = severity_percent(ny_ttl)\nny_sp_sd = severity_percent(ny_sd)\nny_sp_2019 = severity_percent(ny_2019)\nny_sp_2018 = severity_percent(ny_2018)\nny_sp_2017 = severity_percent(ny_2017)\n\n# then again create a simple dataframe for comparison\n\nny_severity_data = pd.DataFrame({'Severity_Rank': [1,2,3,4],\n 'TTL': ny_sp_ttl,\n 'Shut_Down': ny_sp_sd,\n '2019': ny_sp_2019,\n '2018': ny_sp_2018,\n '2017': ny_sp_2017\n })\n\nny_severity_data", "_____no_output_____" ] ], [ [ "#### NY Takeaways:\n- Frequency: See an increase in shutdown period, although there was also an even larger jump from 2018 to 2019\n- Severity: See increase in edge severities (1 & 2)", "_____no_output_____" ], [ "-\n### Pennsylvania\n\nFirst create periods we'll want to compare for the state >", "_____no_output_____" ] ], [ [ "# full state period\npa_ttl = accident_covid[(accident_covid.State == 'PA')]\n\n# shutdown period\npa_sd = accident_covid[\n (accident_covid.State == 'PA') & \n (accident_covid.Shut_Down == 1)]\n\n# 2019 \npa_2019 = accident_covid[(car_accidents_master['Date'] > '2019-03-23') & \n (car_accidents_master['Date'] <= '2019-06-04') & \n (car_accidents_master['State'] == 'PA')]\n\n# 2018\npa_2018 = accident_covid[(car_accidents_master['Date'] > '2018-03-23') & \n (car_accidents_master['Date'] <= '2018-06-04') & \n (car_accidents_master['State'] == 'PA')]\n\n# 2017\npa_2017 = accident_covid[(car_accidents_master['Date'] > '2017-03-23') & \n (car_accidents_master['Date'] <= '2017-06-04') & \n (car_accidents_master['State'] == 'PA')]", "_____no_output_____" ] ], [ [ "#### Frequency\nNow compare frequencies (total number) of car accidents for total, shut-down period in 2020, 2019, 2018, and 2017 (2019, 2018, & 2017 being in same month-day date range as shut-down period) >", "_____no_output_____" ] ], [ [ "# find total frequencies for state total and in each period \n\npa_freqs = [pa_ttl.shape[0],\n pa_sd.shape[0],\n pa_2019.shape[0],\n pa_2018.shape[0],\n pa_2017.shape[0]\n ]\n\n# then develop a simple dataframe to compare them\npa_freq_data = pd.DataFrame({'Timeframe': ['Total','Shut_Down',2019,2018,2017],\n 'Num_Accidents': pa_freqs})\n\npa_freq_data", "_____no_output_____" ], [ "# percent change in frequency in 2020 stay-home compared to average past years\n\nchange = change_freq(pa_freq_data)\nprint(\"Change Frequency: %.2f%%\" % (change))", "Change Frequency: 90.76%\n" ] ], [ [ "#### Severity\nNow compare severities of car accidents for total, shut-down period in 2020, 2019, 2018, and 2017 (2019, 2018 & 2017 being in same month-day date range as shut-down period) >", "_____no_output_____" ] ], [ [ "# first find percentages of severity ranks to total accidents for \n# total and for each period \n\npa_sp_ttl = severity_percent(pa_ttl)\npa_sp_sd = severity_percent(pa_sd)\npa_sp_2019 = severity_percent(pa_2019)\npa_sp_2018 = severity_percent(pa_2018)\npa_sp_2017 = severity_percent(pa_2017)\n\n# then again create a simple dataframe for comparison\n\npa_severity_data = pd.DataFrame({'Severity_Rank': [1,2,3,4],\n 'TTL': pa_sp_ttl,\n 'Shut_Down': pa_sp_sd,\n '2019': pa_sp_2019,\n '2018': pa_sp_2018,\n '2017': pa_sp_2017\n })\n\npa_severity_data", "_____no_output_____" ] ], [ [ "#### PA Takeaways:\n- Frequency: See a fairly significant increase in accidents during shutdown period\n- Severity: See increase in severity 1 and drop in severity 2, but not that large", "_____no_output_____" ], [ "-\n### Illinois\n\nFirst create periods we'll want to compare for the state >", "_____no_output_____" ] ], [ [ "# full state period\nil_ttl = accident_covid[(accident_covid.State == 'IL')]\n\n# shutdown period\nil_sd = accident_covid[\n (accident_covid.State == 'IL') & \n (accident_covid.Shut_Down == 1)]\n\n# 2019 \nil_2019 = accident_covid[(car_accidents_master['Date'] > '2019-03-25') & \n (car_accidents_master['Date'] <= '2019-05-31') & \n (car_accidents_master['State'] == 'IL')]\n\n# 2018\nil_2018 = accident_covid[(car_accidents_master['Date'] > '2018-03-25') & \n (car_accidents_master['Date'] <= '2018-05-31') & \n (car_accidents_master['State'] == 'IL')]\n\n# 2017\nil_2017 = accident_covid[(car_accidents_master['Date'] > '2017-03-25') & \n (car_accidents_master['Date'] <= '2017-05-31') & \n (car_accidents_master['State'] == 'IL')]", "_____no_output_____" ] ], [ [ "#### Frequency\nNow compare frequencies (total number) of car accidents for total, shut-down period in 2020, 2019, 2018, and 2017 (2019, 2018, & 2017 being in same month-day date range as shut-down period) >", "_____no_output_____" ] ], [ [ "# find total frequencies for state total and in each period \n\nil_freqs = [il_ttl.shape[0],\n il_sd.shape[0],\n il_2019.shape[0],\n il_2018.shape[0],\n il_2017.shape[0]\n ]\n\n# then develop a simple dataframe to compare them\nil_freq_data = pd.DataFrame({'Timeframe': ['Total','Shut_Down',2019,2018,2017],\n 'Num_Accidents': il_freqs})\n\nil_freq_data", "_____no_output_____" ], [ "# percent change in frequency in 2020 stay-home compared to average past years\n\nchange = change_freq(il_freq_data)\nprint(\"Change Frequency: %.2f%%\" % (change))", "Change Frequency: 24.55%\n" ] ], [ [ "#### Severity\nNow compare severities of car accidents for total, shut-down period in 2020, 2019, 2018, and 2017 (2019, 2018 & 2017 being in same month-day date range as shut-down period) >", "_____no_output_____" ] ], [ [ "# first find percentages of severity ranks to total accidents for \n# total and for each period \n\nil_sp_ttl = severity_percent(il_ttl)\nil_sp_sd = severity_percent(il_sd)\nil_sp_2019 = severity_percent(il_2019)\nil_sp_2018 = severity_percent(il_2018)\nil_sp_2017 = severity_percent(il_2017)\n\n# then again create a simple dataframe for comparison\n\nil_severity_data = pd.DataFrame({'Severity_Rank': [1,2,3,4],\n 'TTL': il_sp_ttl,\n 'Shut_Down': il_sp_sd,\n '2019': il_sp_2019,\n '2018': il_sp_2018,\n '2017': il_sp_2017\n })\n\nil_severity_data", "_____no_output_____" ] ], [ [ "#### IL Takeaways:\n- Frequency: Increase during shutdown period\n- Severity: Large increase in severity 3 and decrese in severity 2. Also increase in severity 1", "_____no_output_____" ], [ "-\n### Virginia\n\nFirst create periods we'll want to compare for the state >", "_____no_output_____" ] ], [ [ "# full state period\nva_ttl = accident_covid[(accident_covid.State == 'VA')]\n\n# shutdown period\nva_sd = accident_covid[\n (accident_covid.State == 'VA') & \n (accident_covid.Shut_Down == 1)]\n\n# 2019 \nva_2019 = accident_covid[(car_accidents_master['Date'] > '2019-03-30') & \n (car_accidents_master['Date'] <= '2019-06-10') & \n (car_accidents_master['State'] == 'VA')]\n\n# 2018\nva_2018 = accident_covid[(car_accidents_master['Date'] > '2018-03-30') & \n (car_accidents_master['Date'] <= '2018-06-10') & \n (car_accidents_master['State'] == 'VA')]\n\n# 2017\nva_2017 = accident_covid[(car_accidents_master['Date'] > '2017-03-30') & \n (car_accidents_master['Date'] <= '2017-06-10') & \n (car_accidents_master['State'] == 'VA')]", "_____no_output_____" ] ], [ [ "#### Frequency\nNow compare frequencies (total number) of car accidents for total, shut-down period in 2020, 2019, 2018, and 2017 (2019, 2018, & 2017 being in same month-day date range as shut-down period) >", "_____no_output_____" ] ], [ [ "# find total frequencies for state total and in each period \n\nva_freqs = [va_ttl.shape[0],\n va_sd.shape[0],\n va_2019.shape[0],\n va_2018.shape[0],\n va_2017.shape[0]\n ]\n\n# then develop a simple dataframe to compare them\nva_freq_data = pd.DataFrame({'Timeframe': ['Total','Shut_Down',2019,2018,2017],\n 'Num_Accidents': va_freqs})\n\nva_freq_data", "_____no_output_____" ], [ "# percent change in frequency in 2020 stay-home compared to average past years\n\nchange = change_freq(va_freq_data)\nprint(\"Change Frequency: %.2f%%\" % (change))", "Change Frequency: 113.11%\n" ] ], [ [ "#### Severity\nNow compare severities of car accidents for total, shut-down period in 2020, 2019, 2018, and 2017 (2019, 2018 & 2017 being in same month-day date range as shut-down period) >", "_____no_output_____" ] ], [ [ "# first find percentages of severity ranks to total accidents for \n# total and for each period \n\nva_sp_ttl = severity_percent(va_ttl)\nva_sp_sd = severity_percent(va_sd)\nva_sp_2019 = severity_percent(va_2019)\nva_sp_2018 = severity_percent(va_2018)\nva_sp_2017 = severity_percent(va_2017)\n\n# then again create a simple dataframe for comparison\n\nva_severity_data = pd.DataFrame({'Severity_Rank': [1,2,3,4],\n 'TTL': va_sp_ttl,\n 'Shut_Down': va_sp_sd,\n '2019': va_sp_2019,\n '2018': va_sp_2018,\n '2017': va_sp_2017\n })\n\nva_severity_data", "_____no_output_____" ] ], [ [ "#### VA Takeaways:\n- Frequency: See a large increase in frequency during shutdown\n- Severity: Significant increase in severity 1 and drop in severity 3", "_____no_output_____" ], [ "-\n### Michigan\n\nFirst create periods we'll want to compare for the state >", "_____no_output_____" ] ], [ [ "# full state period\nmi_ttl = accident_covid[(accident_covid.State == 'MI')]\n\n# shutdown period\nmi_sd = accident_covid[\n (accident_covid.State == 'MI') & \n (accident_covid.Shut_Down == 1)]\n\n# 2019 \nmi_2019 = accident_covid[(car_accidents_master['Date'] > '2019-03-24') & \n (car_accidents_master['Date'] <= '2019-05-28') & \n (car_accidents_master['State'] == 'MI')]\n\n# 2018\nmi_2018 = accident_covid[(car_accidents_master['Date'] > '2018-03-24') & \n (car_accidents_master['Date'] <= '2018-05-28') & \n (car_accidents_master['State'] == 'MI')]\n\n# 2017\nmi_2017 = accident_covid[(car_accidents_master['Date'] > '2017-03-24') & \n (car_accidents_master['Date'] <= '2017-05-28') & \n (car_accidents_master['State'] == 'MI')]", "_____no_output_____" ] ], [ [ "#### Frequency\nNow compare frequencies (total number) of car accidents for total, shut-down period in 2020, 2019, 2018, and 2017 (2019, 2018, & 2017 being in same month-day date range as shut-down period) >", "_____no_output_____" ] ], [ [ "# find total frequencies for state total and in each period \n\nmi_freqs = [mi_ttl.shape[0],\n mi_sd.shape[0],\n mi_2019.shape[0],\n mi_2018.shape[0],\n mi_2017.shape[0]\n ]\n\n# then develop a simple dataframe to compare them\nmi_freq_data = pd.DataFrame({'Timeframe': ['Total','Shut_Down',2019,2018,2017],\n 'Num_Accidents': mi_freqs})\n\nmi_freq_data", "_____no_output_____" ], [ "# percent change in frequency in 2020 stay-home compared to average past years\n\nchange = change_freq(mi_freq_data)\nprint(\"Change Frequency: %.2f%%\" % (change))", "Change Frequency: -51.62%\n" ] ], [ [ "#### Severity\nNow compare severities of car accidents for total, shut-down period in 2020, 2019, 2018, and 2017 (2019, 2018 & 2017 being in same month-day date range as shut-down period) >", "_____no_output_____" ] ], [ [ "# first find percentages of severity ranks to total accidents for \n# total and for each period \n\nmi_sp_ttl = severity_percent(mi_ttl)\nmi_sp_sd = severity_percent(mi_sd)\nmi_sp_2019 = severity_percent(mi_2019)\nmi_sp_2018 = severity_percent(mi_2018)\nmi_sp_2017 = severity_percent(mi_2017)\n\n# then again create a simple dataframe for comparison\n\nmi_severity_data = pd.DataFrame({'Severity_Rank': [1,2,3,4],\n 'TTL': mi_sp_ttl,\n 'Shut_Down': mi_sp_sd,\n '2019': mi_sp_2019,\n '2018': mi_sp_2018,\n '2017': mi_sp_2017\n })\n\nmi_severity_data", "_____no_output_____" ] ], [ [ "#### MI Takeaways:\n- Frequency: Drop in number of accidents\n- Severity: Increase in edges of severity 1 & 4 -- most dramatic with severity 4", "_____no_output_____" ], [ "-\n### Georgia\n\nFirst create periods we'll want to compare for the state >", "_____no_output_____" ] ], [ [ "# full state period\nga_ttl = accident_covid[(accident_covid.State == 'GA')]\n\n# shutdown period\nga_sd = accident_covid[\n (accident_covid.State == 'GA') & \n (accident_covid.Shut_Down == 1)]\n\n# 2019 \nga_2019 = accident_covid[(car_accidents_master['Date'] > '2019-04-03') & \n (car_accidents_master['Date'] <= '2019-04-30') & \n (car_accidents_master['State'] == 'GA')]\n\n# 2018\nga_2018 = accident_covid[(car_accidents_master['Date'] > '2018-04-03') & \n (car_accidents_master['Date'] <= '2018-04-30') & \n (car_accidents_master['State'] == 'GA')]\n\n# 2017\nga_2017 = accident_covid[(car_accidents_master['Date'] > '2017-04-03') & \n (car_accidents_master['Date'] <= '2017-04-30') & \n (car_accidents_master['State'] == 'GA')]", "_____no_output_____" ] ], [ [ "#### Frequency\nNow compare frequencies (total number) of car accidents for total, shut-down period in 2020, 2019, 2018, and 2017 (2019, 2018, & 2017 being in same month-day date range as shut-down period) >", "_____no_output_____" ] ], [ [ "# find total frequencies for state total and in each period \n\nga_freqs = [ga_ttl.shape[0],\n ga_sd.shape[0],\n ga_2019.shape[0],\n ga_2018.shape[0],\n ga_2017.shape[0]\n ]\n\n# then develop a simple dataframe to compare them\nga_freq_data = pd.DataFrame({'Timeframe': ['Total','Shut_Down',2019,2018,2017],\n 'Num_Accidents': ga_freqs})\n\nga_freq_data", "_____no_output_____" ], [ "# percent change in frequency in 2020 stay-home compared to average past years\n\nchange = change_freq(ga_freq_data)\nprint(\"Change Frequency: %.2f%%\" % (change))", "Change Frequency: -13.71%\n" ] ], [ [ "#### Severity\nNow compare severities of car accidents for total, shut-down period in 2020, 2019, 2018, and 2017 (2019, 2018 & 2017 being in same month-day date range as shut-down period) >", "_____no_output_____" ] ], [ [ "# first find percentages of severity ranks to total accidents for \n# total and for each period \n\nga_sp_ttl = severity_percent(ga_ttl)\nga_sp_sd = severity_percent(ga_sd)\nga_sp_2019 = severity_percent(ga_2019)\nga_sp_2018 = severity_percent(ga_2018)\nga_sp_2017 = severity_percent(ga_2017)\n\n# then again create a simple dataframe for comparison\n\nga_severity_data = pd.DataFrame({'Severity_Rank': [1,2,3,4],\n 'TTL': ga_sp_ttl,\n 'Shut_Down': ga_sp_sd,\n '2019': ga_sp_2019,\n '2018': ga_sp_2018,\n '2017': ga_sp_2017\n })\n\nga_severity_data", "_____no_output_____" ] ], [ [ "-\n### Oregon\n\nFirst create periods we'll want to compare for the state >", "_____no_output_____" ] ], [ [ "# full state period\nor_ttl = accident_covid[(accident_covid.State == 'OR')]\n\n# shutdown period\nor_sd = accident_covid[\n (accident_covid.State == 'OR') & \n (accident_covid.Shut_Down == 1)]\n\n# 2019 \nor_2019 = accident_covid[(car_accidents_master['Date'] > '2019-03-23') & \n (car_accidents_master['Date'] <= '2019-06-30') & \n (car_accidents_master['State'] == 'OR')]\n\n# 2018\nor_2018 = accident_covid[(car_accidents_master['Date'] > '2018-03-23') & \n (car_accidents_master['Date'] <= '2018-06-30') & \n (car_accidents_master['State'] == 'OR')]\n\n# 2017\nor_2017 = accident_covid[(car_accidents_master['Date'] > '2017-03-23') & \n (car_accidents_master['Date'] <= '2017-06-30') & \n (car_accidents_master['State'] == 'OR')]", "_____no_output_____" ] ], [ [ "#### Frequency\nNow compare frequencies (total number) of car accidents for total, shut-down period in 2020, 2019, 2018, and 2017 (2019, 2018, & 2017 being in same month-day date range as shut-down period) >m", "_____no_output_____" ] ], [ [ "# find total frequencies for state total and in each period \n\nor_freqs = [or_ttl.shape[0],\n or_sd.shape[0],\n or_2019.shape[0],\n or_2018.shape[0],\n or_2017.shape[0]\n ]\n\n# then develop a simple dataframe to compare them\nor_freq_data = pd.DataFrame({'Timeframe': ['Total','Shut_Down',2019,2018,2017],\n 'Num_Accidents': or_freqs})\n\nor_freq_data", "_____no_output_____" ], [ "# percent change in frequency in 2020 stay-home compared to average past years\n\nchange = change_freq(or_freq_data)\nprint(\"Change Frequency: %.2f%%\" % (change))", "Change Frequency: 187.12%\n" ] ], [ [ "#### Severity\nNow compare severities of car accidents for total, shut-down period in 2020, 2019, 2018, and 2017 (2019, 2018 & 2017 being in same month-day date range as shut-down period) >", "_____no_output_____" ] ], [ [ "# first find percentages of severity ranks to total accidents for \n# total and for each period \n\nor_sp_ttl = severity_percent(or_ttl)\nor_sp_sd = severity_percent(or_sd)\nor_sp_2019 = severity_percent(or_2019)\nor_sp_2018 = severity_percent(or_2018)\nor_sp_2017 = severity_percent(or_2017)\n\n# then again create a simple dataframe for comparison\n\nor_severity_data = pd.DataFrame({'Severity_Rank': [1,2,3,4],\n 'TTL': or_sp_ttl,\n 'Shut_Down': or_sp_sd,\n '2019': or_sp_2019,\n '2018': or_sp_2018,\n '2017': or_sp_2017\n })\n\nor_severity_data", "_____no_output_____" ] ], [ [ "-\n### Minnesota\n\nFirst create periods we'll want to compare for the state >", "_____no_output_____" ] ], [ [ "# full state period\nmn_ttl = accident_covid[(accident_covid.State == 'MN')]\n\n# shutdown period\nmn_sd = accident_covid[\n (accident_covid.State == 'MN') & \n (accident_covid.Shut_Down == 1)]\n\n# 2019 \nmn_2019 = accident_covid[(car_accidents_master['Date'] > '2019-03-27') & \n (car_accidents_master['Date'] <= '2019-05-17') & \n (car_accidents_master['State'] == 'MN')]\n\n# 2018\nmn_2018 = accident_covid[(car_accidents_master['Date'] > '2018-03-27') & \n (car_accidents_master['Date'] <= '2018-05-17') & \n (car_accidents_master['State'] == 'MN')]\n\n# 2017\nmn_2017 = accident_covid[(car_accidents_master['Date'] > '2017-03-27') & \n (car_accidents_master['Date'] <= '2017-05-17') & \n (car_accidents_master['State'] == 'MN')]", "_____no_output_____" ] ], [ [ "#### Frequency\nNow compare frequencies (total number) of car accidents for total, shut-down period in 2020, 2019, 2018, and 2017 (2019, 2018, & 2017 being in same month-day date range as shut-down period) >", "_____no_output_____" ] ], [ [ "# find total frequencies for state total and in each period \n\nmn_freqs = [mn_ttl.shape[0],\n mn_sd.shape[0],\n mn_2019.shape[0],\n mn_2018.shape[0],\n mn_2017.shape[0]\n ]\n\n# then develop a simple dataframe to compare them\nmn_freq_data = pd.DataFrame({'Timeframe': ['Total','Shut_Down',2019,2018,2017],\n 'Num_Accidents': mn_freqs})\n\nmn_freq_data", "_____no_output_____" ], [ "# percent change in frequency in 2020 stay-home compared to average past years\n\nchange = change_freq(mn_freq_data)\nprint(\"Change Frequency: %.2f%%\" % (change))", "Change Frequency: 83.18%\n" ] ], [ [ "#### Severity\nNow compare severities of car accidents for total, shut-down period in 2020, 2019, 2018, and 2017 (2019, 2018 & 2017 being in same month-day date range as shut-down period) >", "_____no_output_____" ] ], [ [ "# first find percentages of severity ranks to total accidents for \n# total and for each period \n\nmn_sp_ttl = severity_percent(mn_ttl)\nmn_sp_sd = severity_percent(mn_sd)\nmn_sp_2019 = severity_percent(mn_2019)\nmn_sp_2018 = severity_percent(mn_2018)\nmn_sp_2017 = severity_percent(mn_2017)\n\n# then again create a simple dataframe for comparison\n\nmn_severity_data = pd.DataFrame({'Severity_Rank': [1,2,3,4],\n 'TTL': mn_sp_ttl,\n 'Shut_Down': mn_sp_sd,\n '2019': mn_sp_2019,\n '2018': mn_sp_2018,\n '2017': mn_sp_2017\n })\n\nmn_severity_data", "_____no_output_____" ] ], [ [ "-\n### Arizona\n\nFirst create periods we'll want to compare for the state >", "_____no_output_____" ] ], [ [ "# full state period\naz_ttl = accident_covid[(accident_covid.State == 'AZ')]\n\n# shutdown period\naz_sd = accident_covid[\n (accident_covid.State == 'AZ') & \n (accident_covid.Shut_Down == 1)]\n\n# 2019 \naz_2019 = accident_covid[(car_accidents_master['Date'] > '2019-03-31') & \n (car_accidents_master['Date'] <= '2019-05-15') & \n (car_accidents_master['State'] == 'AZ')]\n\n# 2018\naz_2018 = accident_covid[(car_accidents_master['Date'] > '2018-03-31') & \n (car_accidents_master['Date'] <= '2018-05-15') & \n (car_accidents_master['State'] == 'AZ')]\n\n# 2017\naz_2017 = accident_covid[(car_accidents_master['Date'] > '2017-03-31') & \n (car_accidents_master['Date'] <= '2017-05-15') & \n (car_accidents_master['State'] == 'AZ')]", "_____no_output_____" ] ], [ [ "#### Frequency\nNow compare frequencies (total number) of car accidents for total, shut-down period in 2020, 2019, 2018, and 2017 (2019, 2018, & 2017 being in same month-day date range as shut-down period) >", "_____no_output_____" ] ], [ [ "# find total frequencies for state total and in each period \n\naz_freqs = [az_ttl.shape[0],\n az_sd.shape[0],\n az_2019.shape[0],\n az_2018.shape[0],\n az_2017.shape[0]\n ]\n\n# then develop a simple dataframe to compare them\naz_freq_data = pd.DataFrame({'Timeframe': ['Total','Shut_Down',2019,2018,2017],\n 'Num_Accidents': az_freqs})\n\naz_freq_data", "_____no_output_____" ], [ "# percent change in frequency in 2020 stay-home compared to average past years\n\nchange = change_freq(az_freq_data)\nprint(\"Change Frequency: %.2f%%\" % (change))", "Change Frequency: 152.00%\n" ] ], [ [ "#### Severity\nNow compare severities of car accidents for total, shut-down period in 2020, 2019, 2018, and 2017 (2019, 2018 & 2017 being in same month-day date range as shut-down period) >", "_____no_output_____" ] ], [ [ "# first find percentages of severity ranks to total accidents for \n# total and for each period \n\naz_sp_ttl = severity_percent(az_ttl)\naz_sp_sd = severity_percent(az_sd)\naz_sp_2019 = severity_percent(az_2019)\naz_sp_2018 = severity_percent(az_2018)\naz_sp_2017 = severity_percent(az_2017)\n\n# then again create a simple dataframe for comparison\n\naz_severity_data = pd.DataFrame({'Severity_Rank': [1,2,3,4],\n 'TTL': az_sp_ttl,\n 'Shut_Down': az_sp_sd,\n '2019': az_sp_2019,\n '2018': az_sp_2018,\n '2017': az_sp_2017\n })\n\naz_severity_data", "_____no_output_____" ] ], [ [ "-\n### Tennessee\n\nFirst create periods we'll want to compare for the state >", "_____no_output_____" ] ], [ [ "# full state period\ntn_ttl = accident_covid[(accident_covid.State == 'TN')]\n\n# shutdown period\ntn_sd = accident_covid[\n (accident_covid.State == 'TN') & \n (accident_covid.Shut_Down == 1)]\n\n# 2019 \ntn_2019 = accident_covid[(car_accidents_master['Date'] > '2019-03-31') & \n (car_accidents_master['Date'] <= '2019-04-30') & \n (car_accidents_master['State'] == 'TN')]\n\n# 2018\ntn_2018 = accident_covid[(car_accidents_master['Date'] > '2018-03-31') & \n (car_accidents_master['Date'] <= '2018-04-30') & \n (car_accidents_master['State'] == 'TN')]\n\n# 2017\ntn_2017 = accident_covid[(car_accidents_master['Date'] > '2017-03-31') & \n (car_accidents_master['Date'] <= '2017-04-30') & \n (car_accidents_master['State'] == 'TN')]", "_____no_output_____" ] ], [ [ "#### Frequency\nNow compare frequencies (total number) of car accidents for total, shut-down period in 2020, 2019, 2018, and 2017 (2019, 2018, & 2017 being in same month-day date range as shut-down period) >", "_____no_output_____" ] ], [ [ "# find total frequencies for state total and in each period \n\ntn_freqs = [tn_ttl.shape[0],\n tn_sd.shape[0],\n tn_2019.shape[0],\n tn_2018.shape[0],\n tn_2017.shape[0]\n ]\n\n# then develop a simple dataframe to compare them\ntn_freq_data = pd.DataFrame({'Timeframe': ['Total','Shut_Down',2019,2018,2017],\n 'Num_Accidents': tn_freqs})\n\ntn_freq_data", "_____no_output_____" ], [ "# percent change in frequency in 2020 stay-home compared to average past years\n\nchange = change_freq(tn_freq_data)\nprint(\"Change Frequency: %.2f%%\" % (change))", "Change Frequency: 2.54%\n" ] ], [ [ "#### Severity\nNow compare severities of car accidents for total, shut-down period in 2020, 2019, 2018, and 2017 (2019, 2018 & 2017 being in same month-day date range as shut-down period) >", "_____no_output_____" ] ], [ [ "# first find percentages of severity ranks to total accidents for \n# total and for each period \n\ntn_sp_ttl = severity_percent(tn_ttl)\ntn_sp_sd = severity_percent(tn_sd)\ntn_sp_2019 = severity_percent(tn_2019)\ntn_sp_2018 = severity_percent(tn_2018)\ntn_sp_2017 = severity_percent(tn_2017)\n\n# then again create a simple dataframe for comparison\n\ntn_severity_data = pd.DataFrame({'Severity_Rank': [1,2,3,4],\n 'TTL': tn_sp_ttl,\n 'Shut_Down': tn_sp_sd,\n '2019': tn_sp_2019,\n '2018': tn_sp_2018,\n '2017': tn_sp_2017\n })\n\ntn_severity_data", "_____no_output_____" ] ], [ [ "-\n### Washington\n\nFirst create periods we'll want to compare for the state >", "_____no_output_____" ] ], [ [ "# full state period\nwa_ttl = accident_covid[(accident_covid.State == 'WA')]\n\n# shutdown period\nwa_sd = accident_covid[\n (accident_covid.State == 'WA') & \n (accident_covid.Shut_Down == 1)]\n\n# 2019 \nwa_2019 = accident_covid[(car_accidents_master['Date'] > '2019-03-23') & \n (car_accidents_master['Date'] <= '2019-05-31') & \n (car_accidents_master['State'] == 'WA')]\n\n# 2018\nwa_2018 = accident_covid[(car_accidents_master['Date'] > '2018-03-23') & \n (car_accidents_master['Date'] <= '2018-05-31') & \n (car_accidents_master['State'] == 'WA')]\n\n# 2017\nwa_2017 = accident_covid[(car_accidents_master['Date'] > '2017-03-23') & \n (car_accidents_master['Date'] <= '2017-05-31') & \n (car_accidents_master['State'] == 'WA')]", "_____no_output_____" ] ], [ [ "#### Frequency\nNow compare frequencies (total number) of car accidents for total, shut-down period in 2020, 2019, 2018, and 2017 (2019, 2018, & 2017 being in same month-day date range as shut-down period) >", "_____no_output_____" ] ], [ [ "# find total frequencies for state total and in each period \n\nwa_freqs = [wa_ttl.shape[0],\n wa_sd.shape[0],\n wa_2019.shape[0],\n wa_2018.shape[0],\n wa_2017.shape[0]\n ]\n\n# then develop a simple dataframe to compare them\nwa_freq_data = pd.DataFrame({'Timeframe': ['Total','Shut_Down',2019,2018,2017],\n 'Num_Accidents': wa_freqs})\n\nwa_freq_data", "_____no_output_____" ], [ "# percent change in frequency in 2020 stay-home compared to average past years\n\nchange = change_freq(wa_freq_data)\nprint(\"Change Frequency: %.2f%%\" % (change))", "Change Frequency: -16.54%\n" ] ], [ [ "#### Severity\nNow compare severities of car accidents for total, shut-down period in 2020, 2019, 2018, and 2017 (2019, 2018 & 2017 being in same month-day date range as shut-down period) >", "_____no_output_____" ] ], [ [ "# first find percentages of severity ranks to total accidents for \n# total and for each period \n\nwa_sp_ttl = severity_percent(wa_ttl)\nwa_sp_sd = severity_percent(wa_sd)\nwa_sp_2019 = severity_percent(wa_2019)\nwa_sp_2018 = severity_percent(wa_2018)\nwa_sp_2017 = severity_percent(wa_2017)\n\n# then again create a simple dataframe for comparison\n\nwa_severity_data = pd.DataFrame({'Severity_Rank': [1,2,3,4],\n 'TTL': wa_sp_ttl,\n 'Shut_Down': wa_sp_sd,\n '2019': wa_sp_2019,\n '2018': wa_sp_2018,\n '2017': wa_sp_2017\n })\n\nwa_severity_data", "_____no_output_____" ] ], [ [ "-\n### Ohio\n\nFirst create periods we'll want to compare for the state >", "_____no_output_____" ] ], [ [ "# full state period\noh_ttl = accident_covid[(accident_covid.State == 'OH')]\n\n# shutdown period\noh_sd = accident_covid[\n (accident_covid.State == 'OH') & \n (accident_covid.Shut_Down == 1)]\n\n# 2019 \noh_2019 = accident_covid[(car_accidents_master['Date'] > '2019-03-22') & \n (car_accidents_master['Date'] <= '2019-06-13') & \n (car_accidents_master['State'] == 'OH')]\n\n# 2018\noh_2018 = accident_covid[(car_accidents_master['Date'] > '2018-03-22') & \n (car_accidents_master['Date'] <= '2018-06-13') & \n (car_accidents_master['State'] == 'OH')]\n\n# 2017\noh_2017 = accident_covid[(car_accidents_master['Date'] > '2017-03-22') & \n (car_accidents_master['Date'] <= '2017-06-13') & \n (car_accidents_master['State'] == 'OH')]", "_____no_output_____" ] ], [ [ "#### Frequency\nNow compare frequencies (total number) of car accidents for total, shut-down period in 2020, 2019, 2018, and 2017 (2019, 2018, & 2017 being in same month-day date range as shut-down period) >", "_____no_output_____" ] ], [ [ "# find total frequencies for state total and in each period \n\noh_freqs = [oh_ttl.shape[0],\n oh_sd.shape[0],\n oh_2019.shape[0],\n oh_2018.shape[0],\n oh_2017.shape[0]\n ]\n\n# then develop a simple dataframe to compare them\noh_freq_data = pd.DataFrame({'Timeframe': ['Total','Shut_Down',2019,2018,2017],\n 'Num_Accidents': oh_freqs})\n\noh_freq_data", "_____no_output_____" ], [ "# percent change in frequency in 2020 stay-home compared to average past years\n\nchange = change_freq(oh_freq_data)\nprint(\"Change Frequency: %.2f%%\" % (change))", "Change Frequency: 117.88%\n" ] ], [ [ "#### Severity\nNow compare severities of car accidents for total, shut-down period in 2020, 2019, 2018, and 2017 (2019, 2018 & 2017 being in same month-day date range as shut-down period) >", "_____no_output_____" ] ], [ [ "# first find percentages of severity ranks to total accidents for \n# total and for each period \n\noh_sp_ttl = severity_percent(oh_ttl)\noh_sp_sd = severity_percent(oh_sd)\noh_sp_2019 = severity_percent(oh_2019)\noh_sp_2018 = severity_percent(oh_2018)\noh_sp_2017 = severity_percent(oh_2017)\n\n# then again create a simple dataframe for comparison\n\noh_severity_data = pd.DataFrame({'Severity_Rank': [1,2,3,4],\n 'TTL': oh_sp_ttl,\n 'Shut_Down': oh_sp_sd,\n '2019': oh_sp_2019,\n '2018': oh_sp_2018,\n '2017': oh_sp_2017\n })\n\noh_severity_data", "_____no_output_____" ] ], [ [ "-\n### Louisiana\n\nFirst create periods we'll want to compare for the state >", "_____no_output_____" ] ], [ [ "# full state period\nla_ttl = accident_covid[(accident_covid.State == 'LA')]\n\n# shutdown period\nla_sd = accident_covid[\n (accident_covid.State == 'LA') & \n (accident_covid.Shut_Down == 1)]\n\n# 2019 \nla_2019 = accident_covid[(car_accidents_master['Date'] > '2019-03-22') & \n (car_accidents_master['Date'] <= '2019-05-15') & \n (car_accidents_master['State'] == 'LA')]\n\n# 2018\nla_2018 = accident_covid[(car_accidents_master['Date'] > '2018-03-22') & \n (car_accidents_master['Date'] <= '2018-05-15') & \n (car_accidents_master['State'] == 'LA')]\n\n# 2017\nla_2017 = accident_covid[(car_accidents_master['Date'] > '2017-03-22') & \n (car_accidents_master['Date'] <= '2017-05-15') & \n (car_accidents_master['State'] == 'LA')]", "_____no_output_____" ] ], [ [ "#### Frequency\nNow compare frequencies (total number) of car accidents for total, shut-down period in 2020, 2019, 2018, and 2017 (2019, 2018, & 2017 being in same month-day date range as shut-down period) >", "_____no_output_____" ] ], [ [ "# find total frequencies for state total and in each period \n\nla_freqs = [la_ttl.shape[0],\n la_sd.shape[0],\n la_2019.shape[0],\n la_2018.shape[0],\n la_2017.shape[0]\n ]\n\n# then develop a simple dataframe to compare them\nla_freq_data = pd.DataFrame({'Timeframe': ['Total','Shut_Down',2019,2018,2017],\n 'Num_Accidents': la_freqs})\n\nla_freq_data", "_____no_output_____" ], [ "# percent change in frequency in 2020 stay-home compared to average past years\n\nchange = change_freq(la_freq_data)\nprint(\"Change Frequency: %.2f%%\" % (change))", "Change Frequency: 13.45%\n" ] ], [ [ "#### Severity\nNow compare severities of car accidents for total, shut-down period in 2020, 2019, 2018, and 2017 (2019, 2018 & 2017 being in same month-day date range as shut-down period) >", "_____no_output_____" ] ], [ [ "# first find percentages of severity ranks to total accidents for \n# total and for each period \n\nla_sp_ttl = severity_percent(la_ttl)\nla_sp_sd = severity_percent(la_sd)\nla_sp_2019 = severity_percent(la_2019)\nla_sp_2018 = severity_percent(la_2018)\nla_sp_2017 = severity_percent(la_2017)\n\n# then again create a simple dataframe for comparison\n\nla_severity_data = pd.DataFrame({'Severity_Rank': [1,2,3,4],\n 'TTL': la_sp_ttl,\n 'Shut_Down': la_sp_sd,\n '2019': la_sp_2019,\n '2018': la_sp_2018,\n '2017': la_sp_2017\n })\n\nla_severity_data", "_____no_output_____" ] ], [ [ "-\n### Oklahoma\n\nFirst create periods we'll want to compare for the state >", "_____no_output_____" ] ], [ [ "# full state period\nok_ttl = accident_covid[(accident_covid.State == 'OK')]\n\n# shutdown period\nok_sd = accident_covid[\n (accident_covid.State == 'OK') & \n (accident_covid.Shut_Down == 1)]\n\n# 2019 \nok_2019 = accident_covid[(car_accidents_master['Date'] > '2019-03-24') & \n (car_accidents_master['Date'] <= '2019-05-06') & \n (car_accidents_master['State'] == 'OK')]\n\n# 2018\nok_2018 = accident_covid[(car_accidents_master['Date'] > '2018-03-24') & \n (car_accidents_master['Date'] <= '2018-05-06') & \n (car_accidents_master['State'] == 'OK')]\n\n# 2017\nok_2017 = accident_covid[(car_accidents_master['Date'] > '2017-03-24') & \n (car_accidents_master['Date'] <= '2017-05-06') & \n (car_accidents_master['State'] == 'OK')]", "_____no_output_____" ] ], [ [ "#### Frequency\nNow compare frequencies (total number) of car accidents for total, shut-down period in 2020, 2019, 2018, and 2017 (2019, 2018, & 2017 being in same month-day date range as shut-down period) >", "_____no_output_____" ] ], [ [ "# find total frequencies for state total and in each period \n\nok_freqs = [ok_ttl.shape[0],\n ok_sd.shape[0],\n ok_2019.shape[0],\n ok_2018.shape[0],\n ok_2017.shape[0]\n ]\n\n# then develop a simple dataframe to compare them\nok_freq_data = pd.DataFrame({'Timeframe': ['Total','Shut_Down',2019,2018,2017],\n 'Num_Accidents': ok_freqs})\n\nok_freq_data", "_____no_output_____" ], [ "# percent change in frequency in 2020 stay-home compared to average past years\n\nchange = change_freq(ok_freq_data)\nprint(\"Change Frequency: %.2f%%\" % (change))", "Change Frequency: 4.37%\n" ] ], [ [ "#### Severity\nNow compare severities of car accidents for total, shut-down period in 2020, 2019, 2018, and 2017 (2019, 2018 & 2017 being in same month-day date range as shut-down period) >", "_____no_output_____" ] ], [ [ "# first find percentages of severity ranks to total accidents for \n# total and for each period \n\nok_sp_ttl = severity_percent(ok_ttl)\nok_sp_sd = severity_percent(ok_sd)\nok_sp_2019 = severity_percent(ok_2019)\nok_sp_2018 = severity_percent(ok_2018)\nok_sp_2017 = severity_percent(ok_2017)\n\n# then again create a simple dataframe for comparison\n\nok_severity_data = pd.DataFrame({'Severity_Rank': [1,2,3,4],\n 'TTL': ok_sp_ttl,\n 'Shut_Down': ok_sp_sd,\n '2019': ok_sp_2019,\n '2018': ok_sp_2018,\n '2017': ok_sp_2017\n })\n\nok_severity_data", "_____no_output_____" ] ], [ [ "-\n### New Jersey\n\nFirst create periods we'll want to compare for the state >", "_____no_output_____" ] ], [ [ "# full state period\nnj_ttl = accident_covid[(accident_covid.State == 'NJ')]\n\n# shutdown period\nnj_sd = accident_covid[\n (accident_covid.State == 'NJ') & \n (accident_covid.Shut_Down == 1)]\n\n# 2019 \nnj_2019 = accident_covid[(car_accidents_master['Date'] > '2019-03-21') & \n (car_accidents_master['Date'] <= '2019-06-30') & \n (car_accidents_master['State'] == 'NJ')]\n\n# 2018\nnj_2018 = accident_covid[(car_accidents_master['Date'] > '2018-03-21') & \n (car_accidents_master['Date'] <= '2018-06-30') & \n (car_accidents_master['State'] == 'NJ')]\n\n# 2017\nnj_2017 = accident_covid[(car_accidents_master['Date'] > '2017-03-21') & \n (car_accidents_master['Date'] <= '2017-06-30') & \n (car_accidents_master['State'] == 'NJ')]", "_____no_output_____" ] ], [ [ "#### Frequency\nNow compare frequencies (total number) of car accidents for total, shut-down period in 2020, 2019, 2018, and 2017 (2019, 2018, & 2017 being in same month-day date range as shut-down period) >", "_____no_output_____" ] ], [ [ "# find total frequencies for state total and in each period \n\nnj_freqs = [nj_ttl.shape[0],\n nj_sd.shape[0],\n nj_2019.shape[0],\n nj_2018.shape[0],\n nj_2017.shape[0]\n ]\n\n# then develop a simple dataframe to compare them\nnj_freq_data = pd.DataFrame({'Timeframe': ['Total','Shut_Down',2019,2018,2017],\n 'Num_Accidents': nj_freqs})\n\nnj_freq_data", "_____no_output_____" ], [ "# percent change in frequency in 2020 stay-home compared to average past years\n\nchange = change_freq(nj_freq_data)\nprint(\"Change Frequency: %.2f%%\" % (change))", "Change Frequency: 18.74%\n" ] ], [ [ "#### Severity\nNow compare severities of car accidents for total, shut-down period in 2020, 2019, 2018, and 2017 (2019, 2018 & 2017 being in same month-day date range as shut-down period) >", "_____no_output_____" ] ], [ [ "# first find percentages of severity ranks to total accidents for \n# total and for each period \n\nnj_sp_ttl = severity_percent(nj_ttl)\nnj_sp_sd = severity_percent(nj_sd)\nnj_sp_2019 = severity_percent(nj_2019)\nnj_sp_2018 = severity_percent(nj_2018)\nnj_sp_2017 = severity_percent(nj_2017)\n\n# then again create a simple dataframe for comparison\n\nnj_severity_data = pd.DataFrame({'Severity_Rank': [1,2,3,4],\n 'TTL': nj_sp_ttl,\n 'Shut_Down': nj_sp_sd,\n '2019': nj_sp_2019,\n '2018': nj_sp_2018,\n '2017': nj_sp_2017\n })\n\nnj_severity_data", "_____no_output_____" ] ], [ [ "-\n### Top 20 States Averages\n\nCombining numbers for the Top 20 states to see average trends for those states >", "_____no_output_____" ], [ "#### Frequency", "_____no_output_____" ] ], [ [ "# combine fequency data for the top 20 states \n\ntop20_freq_data = (ca_freq_data + \n tx_freq_data +\n fl_freq_data +\n sc_freq_data +\n nc_freq_data +\n ny_freq_data +\n pa_freq_data +\n il_freq_data +\n va_freq_data +\n mi_freq_data +\n ga_freq_data + \n or_freq_data +\n mn_freq_data +\n az_freq_data +\n tn_freq_data +\n wa_freq_data +\n oh_freq_data +\n la_freq_data +\n ok_freq_data +\n nj_freq_data)\n\ntop20_freq_data['Timeframe'] = ['Total','Shut_Down',2019,2018,2017]\n\ntop20_freq_data", "_____no_output_____" ] ], [ [ "Now find change in 2020 shutdown period compared to average of past three years >", "_____no_output_____" ] ], [ [ "# percent change in frequency in 2020 stay-home compared to average past years\n\nchange = change_freq(top20_freq_data)\nprint(\"Change Frequency: %.2f%%\" % (change))", "Change Frequency: 51.77%\n" ] ], [ [ "#### Severity\n\nFor average severity rank for top 10 states, we will include weighting for each state based on the states number of accidents compared to all accidents for the top 10 states", "_____no_output_____" ] ], [ [ "# finding weight for each state to then include in next cell\n\nlen(accident_covid[accident_covid.State == 'NJ'])/accident_covid.State.value_counts()[:20].sum()", "_____no_output_____" ], [ "# add all state severity dfs together\n\ntop20_severity_data = (ca_severity_data * 27 +\n tx_severity_data * 11 +\n fl_severity_data * 9 +\n sc_severity_data * 6 +\n nc_severity_data * 5 +\n ny_severity_data * 5 +\n pa_severity_data * 4 +\n il_severity_data * 3 +\n va_severity_data * 3 +\n mi_severity_data * 3 +\n ga_severity_data * 3 +\n or_severity_data * 3 +\n mn_severity_data * 3 +\n az_severity_data * 3 +\n tn_severity_data * 2 +\n wa_severity_data * 2 +\n oh_severity_data * 2 +\n la_severity_data * 2 +\n ok_severity_data * 2 +\n nj_severity_data * 2 )\n\n# get average for the 20 states\ntop20_severity_data = round(top20_severity_data.div(104),2)\n\ntop20_severity_data", "_____no_output_____" ], [ "# (2020 value - average 2017-2019) / average 2017-2019 * 100\n\nsev1_2020 = top20_severity_data.Shut_Down[0]\n\nsev1_avg_past = np.mean(top20_severity_data['2017'][0] + \n top20_severity_data['2018'][0] + \n top20_severity_data['2019'][0])\n\nprint('Percent Change in 2020 stay-home period vs. average 2017-2019:')\n((sev1_2020 - sev1_avg_past) / sev1_avg_past) * 100", "Percent Change in 2020 stay-home period vs. average 2017-2019:\n" ] ], [ [ "#### Top 20 Takeaways:\n- Frequency: See a 51% increase during shutdown period compared to past periods\n- Severity: See greatest shift in severity 1 -- an increase of 8,870%. Severity 4's are also showing an increase compared to 2019 and 2018, but not to 2017. Changes in severity 4 seem to be more on a state-by-state basis", "_____no_output_____" ] ] ]

[ "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown" ]

[ [ "markdown", "markdown", "markdown" ], [ "code" ], [ "markdown", "markdown" ], [ "code" ], [ "markdown" ], [ "code" ], [ "markdown" ], [ "code" ], [ "markdown" ], [ "code", "code" ], [ "markdown" ], [ "code", "code" ], [ "markdown", "markdown", "markdown" ], [ "code" ], [ "markdown" ], [ "code", "code" ], [ "markdown" ], [ "code" ], [ "markdown", "markdown", "markdown" ], [ "code" ], [ "markdown" ], [ "code", "code" ], [ "markdown" ], [ "code" ], [ "markdown", "markdown" ], [ "code" ], [ "markdown" ], [ "code", "code" ], [ "markdown" ], [ "code" ], [ "markdown", "markdown" ], [ "code" ], [ "markdown" ], [ "code", "code" ], [ "markdown" ], [ "code" ], [ "markdown", "markdown" ], [ "code" ], [ "markdown" ], [ "code", "code" ], [ "markdown" ], [ "code" ], [ "markdown", "markdown" ], [ "code" ], [ "markdown" ], [ "code", "code" ], [ "markdown" ], [ "code" ], [ "markdown", "markdown" ], [ "code" ], [ "markdown" ], [ "code", "code" ], [ "markdown" ], [ "code" ], [ "markdown", "markdown" ], [ "code" ], [ "markdown" ], [ "code", "code" ], [ "markdown" ], [ "code" ], [ "markdown", "markdown" ], [ "code" ], [ "markdown" ], [ "code", "code" ], [ "markdown" ], [ "code" ], [ "markdown", "markdown" ], [ "code" ], [ "markdown" ], [ "code", "code" ], [ "markdown" ], [ "code" ], [ "markdown", "markdown" ], [ "code" ], [ "markdown" ], [ "code", "code" ], [ "markdown" ], [ "code" ], [ "markdown" ], [ "code" ], [ "markdown" ], [ "code", "code" ], [ "markdown" ], [ "code" ], [ "markdown" ], [ "code" ], [ "markdown" ], [ "code", "code" ], [ "markdown" ], [ "code" ], [ "markdown" ], [ "code" ], [ "markdown" ], [ "code", "code" ], [ "markdown" ], [ "code" ], [ "markdown" ], [ "code" ], [ "markdown" ], [ "code", "code" ], [ "markdown" ], [ "code" ], [ "markdown" ], [ "code" ], [ "markdown" ], [ "code", "code" ], [ "markdown" ], [ "code" ], [ "markdown" ], [ "code" ], [ "markdown" ], [ "code", "code" ], [ "markdown" ], [ "code" ], [ "markdown" ], [ "code" ], [ "markdown" ], [ "code", "code" ], [ "markdown" ], [ "code" ], [ "markdown" ], [ "code" ], [ "markdown" ], [ "code", "code" ], [ "markdown" ], [ "code" ], [ "markdown" ], [ "code" ], [ "markdown" ], [ "code", "code" ], [ "markdown" ], [ "code" ], [ "markdown", "markdown" ], [ "code" ], [ "markdown" ], [ "code" ], [ "markdown" ], [ "code", "code", "code" ], [ "markdown" ] ]

[ "MIT" ]

[ "MIT" ]

[ "MIT" ]

[ [ [ "<img src=\"images/usm.jpg\" width=\"480\" height=\"240\" align=\"left\"/>", "_____no_output_____" ], [ "# MAT281 - Modelos de clasificación", "_____no_output_____" ], [ "## Objetivos de la clase\n\n* Aprender conceptos básicos de los modelos de clasificación en python.", "_____no_output_____" ], [ "## Contenidos\n\n* [Modelos de clasificación](#c1)\n* [Ejemplos con python](#c2)\n", "_____no_output_____" ], [ "## I.- Modelos de clasificación\n\nLos modelos de clasificacion son ocupadas para predecir valores categóricos, por ejemplo, determinar la especie de una flor basado en el largo (y ancho) de su pétalo (y sépalo).\nDentro de los modelos de clasificación, el modelo más básico (y no por eso menos importante) es el **modelo de regresión logística**.\n\n\n### Regresión logística\n\nLa **regresión logística** puede usarse para tratar de correlacionar la probabilidad de una variable cualitativa binaria (asumiremos que puede tomar los valores reales **0** y **1**) con una variable escalar $x$. \n\nLa idea es que la regresión logística aproxime la probabilidad de obtener **0** (no ocurre cierto suceso) o **1** (ocurre el suceso) con el valor de la variable explicativa $x$. \n\nEn esas condiciones, la probabilidad aproximada del suceso se aproximará mediante una función logística del tipo:\n\n$$\\pi(x) =\\dfrac{e^{\\beta_0+\\beta_1x}}{e^{\\beta_0+\\beta_1x}+1}=\\dfrac{1}{e^{-(\\beta_0+\\beta_1x)}+1}$$\n\nque puede reducirse al cálculo de una regresión lineal para la función logit de la probabilidad:\n\n$$g(x) = ln(\\dfrac{\\pi(x)}{1- \\pi(x)})=\\beta_0+\\beta_1 x$$\n\nEs decir, el problema de regresión logística puede ser visto como un problema de regresión lineal.\n\n<img src=\"http://res.cloudinary.com/dyd911kmh/image/upload/f_auto,q_auto:best/v1534281070/linear_vs_logistic_regression_edxw03.png\" width=\"560\" height=\"480\" align=\"center\"/>", "_____no_output_____" ], [ "### ¿ Cómo se mide el error para elos modelos de clasificación?\n\nLos modelos de clasificacion son ocupadas para predecir valores categóricos, por ejemplo, determinar la especie de una flor basado en el largo (y ancho) de su pétalo (y sépalo).Para este caso, es necesario introducir el concepto de matriz de confusión.\n\nLa **matriz de confusión** es una herramienta que permite la visualización del desempeño de un algoritmo \nPara la clasificación de dos clases (por ejemplo, 0 y 1), se tiene la siguiente matriz de confusión:\n\n<img src=\"https://static.tildacdn.com/tild6630-3965-4833-b932-646530343464/9.svg\n\" width=\"480\" height=\"360\" align=\"rigt\"/>\n\nAcá se define:\n\n* **TP = Verdadero positivo**: el modelo predijo la clase positiva correctamente, para ser una clase positiva.\n* **FP = Falso positivo**: el modelo predijo la clase negativa incorrectamente, para ser una clase positiva.\n* **FN = Falso negativo**: el modelo predijo incorrectamente que la clase positiva sería la clase negativa.\n* **TN = Verdadero negativo**: el modelo predijo la clase negativa correctamente, para ser la clase negativa.\n\nEn este contexto, los valores TP Y TN muestran los valores correctos que tuve al momento de realizar la predicción, mientras que los valores de de FN Y FP denotan los valores que me equivoque de clase.\n\nLos conceptos de FN y FP se pueden interpretar con la siguiente imagen:\n\n<img src=\"https://static.tildacdn.com/tild6436-6637-4562-b738-613433303838/error.jpg\n\" width=\"480\" height=\"360\" align=\"rigt\"/>\n\n\nEn este contexto, se busca maximizar el número al máximo la suma de los elementos TP Y TN, mientras que se busca disminuir la suma de los elementos de FN y FP. Para esto se definen las siguientes métricas:\n\n\n1. **Accuracy**\n\n$$accuracy(y,\\hat{y}) = \\dfrac{TP+TN}{TP+TN+FP+FN}$$\n\n2. **Recall**:\n\n$$recall(y,\\hat{y}) = \\dfrac{TP}{TP+FN}$$\n\n3. **Precision**:\n\n$$precision(y,\\hat{y}) = \\dfrac{TP}{TP+FP} $$\n\n4. **F-score**:\n\n$$fscore(y,\\hat{y}) = 2\\times \\dfrac{precision(y,\\hat{y})\\times recall(y,\\hat{y})}{precision(y,\\hat{y})+recall(y,\\hat{y})} $$", "_____no_output_____" ], [ "## II.- Ejemplos con python", "_____no_output_____" ], [ "### a) Dataset Iris (regresión logística)\n\nVeamos un pequeño ejemplo de como se implementa en python. En este ejemplo voy a utilizar el dataset **Iris** que ya viene junto con Scikit-learn y es ideal para practicar con regresiones logística ; el mismo contiene los tipos de flores basado en en largo y ancho de su sépalo y pétalo.", "_____no_output_____" ] ], [ [ "# librerias\n \nimport os\nimport numpy as np\nimport pandas as pd\n\nfrom sklearn import datasets\nfrom sklearn.model_selection import train_test_split\n\nimport matplotlib.pyplot as plt\nimport seaborn as sns \npd.set_option('display.max_columns', 500) # Ver más columnas de los dataframes\n\n# Ver gráficos de matplotlib en jupyter notebook/lab\n%matplotlib inline", "_____no_output_____" ], [ "# cargar datos\niris = datasets.load_iris()\nprint(iris.DESCR)", ".. _iris_dataset:\n\nIris plants dataset\n--------------------\n\n**Data Set Characteristics:**\n\n :Number of Instances: 150 (50 in each of three classes)\n :Number of Attributes: 4 numeric, predictive attributes and the class\n :Attribute Information:\n - sepal length in cm\n - sepal width in cm\n - petal length in cm\n - petal width in cm\n - class:\n - Iris-Setosa\n - Iris-Versicolour\n - Iris-Virginica\n \n :Summary Statistics:\n\n ============== ==== ==== ======= ===== ====================\n Min Max Mean SD Class Correlation\n ============== ==== ==== ======= ===== ====================\n sepal length: 4.3 7.9 5.84 0.83 0.7826\n sepal width: 2.0 4.4 3.05 0.43 -0.4194\n petal length: 1.0 6.9 3.76 1.76 0.9490 (high!)\n petal width: 0.1 2.5 1.20 0.76 0.9565 (high!)\n ============== ==== ==== ======= ===== ====================\n\n :Missing Attribute Values: None\n :Class Distribution: 33.3% for each of 3 classes.\n :Creator: R.A. Fisher\n :Donor: Michael Marshall (MARSHALL%[email protected])\n :Date: July, 1988\n\nThe famous Iris database, first used by Sir R.A. Fisher. The dataset is taken\nfrom Fisher's paper. Note that it's the same as in R, but not as in the UCI\nMachine Learning Repository, which has two wrong data points.\n\nThis is perhaps the best known database to be found in the\npattern recognition literature. Fisher's paper is a classic in the field and\nis referenced frequently to this day. (See Duda & Hart, for example.) The\ndata set contains 3 classes of 50 instances each, where each class refers to a\ntype of iris plant. One class is linearly separable from the other 2; the\nlatter are NOT linearly separable from each other.\n\n.. topic:: References\n\n - Fisher, R.A. \"The use of multiple measurements in taxonomic problems\"\n Annual Eugenics, 7, Part II, 179-188 (1936); also in \"Contributions to\n Mathematical Statistics\" (John Wiley, NY, 1950).\n - Duda, R.O., & Hart, P.E. (1973) Pattern Classification and Scene Analysis.\n (Q327.D83) John Wiley & Sons. ISBN 0-471-22361-1. See page 218.\n - Dasarathy, B.V. (1980) \"Nosing Around the Neighborhood: A New System\n Structure and Classification Rule for Recognition in Partially Exposed\n Environments\". IEEE Transactions on Pattern Analysis and Machine\n Intelligence, Vol. PAMI-2, No. 1, 67-71.\n - Gates, G.W. (1972) \"The Reduced Nearest Neighbor Rule\". IEEE Transactions\n on Information Theory, May 1972, 431-433.\n - See also: 1988 MLC Proceedings, 54-64. Cheeseman et al\"s AUTOCLASS II\n conceptual clustering system finds 3 classes in the data.\n - Many, many more ...\n" ], [ "# dejar en formato dataframe\n\niris_df = pd.DataFrame(iris.data, columns=iris.feature_names)\niris_df['TARGET'] = iris.target\niris_df.head() # estructura de nuestro dataset.", "_____no_output_____" ] ], [ [ "Para ver gráficamente el modelo de regresión logística, ajustemos el modelo solo a dos variables: petal length (cm), petal width (cm).", "_____no_output_____" ] ], [ [ "# datos \nfrom sklearn.linear_model import LogisticRegression\n\nX = iris_df[['sepal length (cm)', 'sepal width (cm)']]\nY = iris_df['TARGET']\n\n# split dataset\nX_train, X_test, Y_train, Y_test = train_test_split(X, Y, test_size=0.2, random_state = 2) ", "_____no_output_____" ], [ "# print rows train and test sets\nprint('Separando informacion:\\n')\nprint('numero de filas data original : ',len(X))\nprint('numero de filas train set : ',len(X_train))\nprint('numero de filas test set : ',len(X_test))", "Separando informacion:\n\nnumero de filas data original : 150\nnumero de filas train set : 120\nnumero de filas test set : 30\n" ], [ "# Creando el modelo\nrlog = LogisticRegression()\nrlog.fit(X_train, Y_train) # ajustando el modelo", "_____no_output_____" ] ], [ [ "Grafiquemos nuestro resultados:\n", "_____no_output_____" ] ], [ [ "# grafica de la regresion logistica \nplt.figure(figsize=(12,4))\n\n# dataframe a matriz\nX = X.values\nY = Y.values\n\nx_min, x_max = X[:, 0].min() - .5, X[:, 0].max() + .5\ny_min, y_max = X[:, 1].min() - .5, X[:, 1].max() + .5\n\nh = .02 # step size in the mesh\nxx, yy = np.meshgrid(np.arange(x_min, x_max, h), np.arange(y_min, y_max, h))\nZ = rlog.predict(np.c_[xx.ravel(), yy.ravel()])\n\n# Put the result into a color plot\nZ = Z.reshape(xx.shape)\nplt.figure(1, figsize=(4, 3))\nplt.pcolormesh(xx, yy, Z, cmap=plt.cm.Paired)\n\n# Plot also the training points\nplt.scatter(X[:, 0], X[:, 1], c=Y, edgecolors='k', cmap=plt.cm.Paired)\nplt.xlabel('Sepal length')\nplt.ylabel('Sepal width')\nplt.show()", "_____no_output_____" ] ], [ [ "Gráficamente podemos decir que el modelo se ajusta bastante bien, puesto que las clasificaciones son adecuadas y el modelo no se confunde entre una clase y otra. Por otro lado, existe valores numéricos que también nos pueden ayudar a convensernos de estos, que son las métricas que se habian definidos con anterioridad. \n\nPara ello, instanciaremos las distintas metricas del archivo **metrics_classification.py** y calcularemos sus distintos valores.", "_____no_output_____" ] ], [ [ "# metrics\n\nfrom metrics_classification import *\nfrom sklearn.metrics import confusion_matrix\n\ny_true = list(Y_test)\ny_pred = list(rlog.predict(X_test))\n\nprint('Valores:\\n')\nprint('originales: ', y_true)\nprint('predicho: ', y_pred)\n\n\nprint('\\nMatriz de confusion:\\n ')\nprint(confusion_matrix(y_true,y_pred))\n\n# ejemplo \ndf_temp = pd.DataFrame(\n {\n 'y':y_true,\n 'yhat':y_pred\n }\n)\n\ndf_metrics = summary_metrics(df_temp)\nprint(\"\\nMetricas para los regresores : 'sepal length (cm)' y 'sepal width (cm)'\")\nprint(\"\")\nprint(df_metrics)", "Valores:\n\noriginales: [0, 0, 2, 0, 0, 2, 0, 2, 2, 0, 0, 0, 0, 0, 1, 1, 0, 1, 2, 1, 1, 1, 2, 1, 1, 0, 0, 2, 0, 2]\npredicho: [0, 0, 1, 0, 0, 2, 0, 2, 2, 0, 0, 0, 0, 0, 2, 2, 1, 1, 2, 1, 2, 2, 2, 1, 1, 0, 0, 1, 0, 2]\n\nMatriz de confusion:\n \n[[13 1 0]\n [ 0 4 4]\n [ 0 2 6]]\n\nMetricas para los regresores : 'sepal length (cm)' y 'sepal width (cm)'\n\n accuracy recall precision fscore\n0 0.7667 0.7262 0.7238 0.721\n" ] ], [ [ "Basado en las métricas y en la gráfica, podemos concluir que el ajuste realizado es bastante asertado. \n\nOtra forma de convencernos que nuestro ajuste es correcto es analizando la [curva AUC–ROC](https://en.wikipedia.org/wiki/Receiver_operating_characteristic).\n\nLa curva AUC-ROC es la métrica de selección del modelo para el problema de clasificación bi-multi class. ROC es una curva de probabilidad para diferentes clases. ROC nos dice qué tan bueno es el modelo para distinguir las clases dadas, en términos de la probabilidad predicha.\n\nUna curva ROC típica tiene una tasa de falsos positivos (FPR) en el eje X y una tasa de verdaderos positivos (TPR) en el eje Y.\n\n<img src=\"https://s3.amazonaws.com/stackabuse/media/understanding-roc-curves-python-3.png\" width=\"480\" height=\"480\" align=\"center\"/>", "_____no_output_____" ], [ "El área cubierta por la curva es el área entre la línea naranja (ROC) y el eje. Esta área cubierta es AUC. Cuanto más grande sea el área cubierta, mejores serán los modelos de aprendizaje automático para distinguir las clases dadas. El valor ideal para AUC es 1.\n\nBasado en este concepto, calculemos la curva AUC-ROC para nuestro ejemplo. Cabe destacar que esta curva es efectiva solo para clasificación binaria, por lo que para efectos prácticos convertiremos nuestro **TARGET** en binarios (0 ó 1).\n\nPara efectos prácticos tranformaremos la clase objetivo (en este caso, la clase **0**) a **1**, y el resto de las clases (clase **1** y **2**) las dejaremos en la clase **0**.", "_____no_output_____" ] ], [ [ "from sklearn.metrics import roc_curve\nfrom sklearn.metrics import roc_auc_score", "_____no_output_____" ], [ "# graficar curva roc\ndef plot_roc_curve(fpr, tpr):\n plt.figure(figsize=(9,4))\n plt.plot(fpr, tpr, color='orange', label='ROC')\n plt.plot([0, 1], [0, 1], color='darkblue', linestyle='--')\n plt.xlabel('False Positive Rate')\n plt.ylabel('True Positive Rate')\n plt.title('Receiver Operating Characteristic (ROC) Curve')\n plt.legend()\n plt.show()", "_____no_output_____" ], [ "# separar clase 0 del resto\nX = iris_df[['sepal length (cm)', 'sepal width (cm)']]\nY = iris_df['TARGET'].apply(lambda x: 1 if x ==2 else 0)\nmodel = LogisticRegression()\n\n# split dataset\nX_train, X_test, Y_train, Y_test = train_test_split(X, Y, test_size=0.3, random_state = 2)\n\n# ajustar modelo \nmodel.fit(X_train,Y_train)", "_____no_output_____" ], [ "# calcular score AUC\n\nprobs = model.predict_proba(X_test) # predecir probabilidades para X_test\nprobs_tp = probs[:, 1] # mantener solo las probabilidades de la clase positiva \n \nauc = roc_auc_score(Y_test, probs_tp) # calcular score AUC \n\nprint('AUC: %.2f' % auc)", "AUC: 0.93\n" ], [ "# calcular curva ROC\nfpr, tpr, thresholds = roc_curve(Y_test, probs_tp) # obtener curva ROC\nplot_roc_curve(fpr, tpr)", "_____no_output_____" ] ], [ [ "### b) Varios modelos de clasificación\n\nExisten varios modelos de clasificación que podemos ir comparando unos con otros, dentro de los cuales estacamos los siguientes:\n\n\n* [Regresión Logística](https://es.wikipedia.org/wiki/Regresi%C3%B3n_log%C3%ADstica)\n* [Arboles de Decision](https://es.wikipedia.org/wiki/%C3%81rbol_de_decisi%C3%B3n)\n* [Random Forest](https://es.wikipedia.org/wiki/Random_forest)\n* [SVM](https://es.wikipedia.org/wiki/M%C3%A1quinas_de_vectores_de_soporte) \n\nNos basaremos en un ejemplo de **sklearn** que muestra los resultados de aplicar estos cuatro modelos sobre tres conjunto de datos distintos ( **make_moons**, **make_circles**, **make_classification**). Además, se crea un rutina para comparar los resultados de las distintas métricas.", "_____no_output_____" ], [ "### a) Gráficos\n\nSimilar al gráfico aplicado al conjunto de datos Iris, aca se realiza el mismo ejercicio pero para tres conjunto de datos sobre los distintos modelos.", "_____no_output_____" ] ], [ [ "from sklearn.datasets import make_moons, make_circles, make_classification\nfrom sklearn.preprocessing import StandardScaler\n\n\nfrom sklearn.linear_model import LogisticRegression\nfrom sklearn.svm import SVC\nfrom sklearn.tree import DecisionTreeClassifier\nfrom sklearn.ensemble import RandomForestClassifier\n\nfrom matplotlib.colors import ListedColormap\n\n\nh = .02 # step size in the mesh\n\nnames = [\"Logistic\",\n \"RBF SVM\", \n \"Decision Tree\", \n \"Random Forest\"\n]\n\nclassifiers = [\n LogisticRegression(),\n SVC(gamma=2, C=1),\n DecisionTreeClassifier(max_depth=5),\n RandomForestClassifier(max_depth=5, n_estimators=10, max_features=1),\n]\n\n\n\nX, y = make_classification(n_features=2, n_redundant=0, n_informative=2,\n random_state=1, n_clusters_per_class=1)\nrng = np.random.RandomState(2)\nX += 2 * rng.uniform(size=X.shape)\nlinearly_separable = (X, y)\n\ndatasets = [make_moons(noise=0.3, random_state=0),\n make_circles(noise=0.2, factor=0.5, random_state=1),\n linearly_separable\n ]\n\n\n\nfigure = plt.figure(figsize=(27, 9))\ni = 1\n\n\n\n# iterate over datasets\nfor ds_cnt, ds in enumerate(datasets):\n # preprocess dataset, split into training and test part\n X, y = ds\n X = StandardScaler().fit_transform(X)\n X_train, X_test, y_train, y_test = \\\n train_test_split(X, y, test_size=.4, random_state=42)\n\n x_min, x_max = X[:, 0].min() - .5, X[:, 0].max() + .5\n y_min, y_max = X[:, 1].min() - .5, X[:, 1].max() + .5\n xx, yy = np.meshgrid(np.arange(x_min, x_max, h),\n np.arange(y_min, y_max, h))\n\n # just plot the dataset first\n cm = plt.cm.RdBu\n cm_bright = ListedColormap(['#FF0000', '#0000FF'])\n ax = plt.subplot(len(datasets), len(classifiers) + 1, i)\n if ds_cnt == 0:\n ax.set_title(\"Input data\")\n # Plot the training points\n ax.scatter(X_train[:, 0], X_train[:, 1], c=y_train, cmap=cm_bright,\n edgecolors='k')\n # Plot the testing points\n ax.scatter(X_test[:, 0], X_test[:, 1], c=y_test, cmap=cm_bright, alpha=0.6,\n edgecolors='k')\n ax.set_xlim(xx.min(), xx.max())\n ax.set_ylim(yy.min(), yy.max())\n ax.set_xticks(())\n ax.set_yticks(())\n i += 1\n\n # iterate over classifiers\n for name, clf in zip(names, classifiers):\n ax = plt.subplot(len(datasets), len(classifiers) + 1, i)\n clf.fit(X_train, y_train)\n score = clf.score(X_test, y_test)\n\n # Plot the decision boundary. For that, we will assign a color to each\n # point in the mesh [x_min, x_max]x[y_min, y_max].\n if hasattr(clf, \"decision_function\"):\n Z = clf.decision_function(np.c_[xx.ravel(), yy.ravel()])\n else:\n Z = clf.predict_proba(np.c_[xx.ravel(), yy.ravel()])[:, 1]\n\n # Put the result into a color plot\n Z = Z.reshape(xx.shape)\n ax.contourf(xx, yy, Z, cmap=cm, alpha=.8)\n\n # Plot the training points\n ax.scatter(X_train[:, 0], X_train[:, 1], c=y_train, cmap=cm_bright,\n edgecolors='k')\n # Plot the testing points\n ax.scatter(X_test[:, 0], X_test[:, 1], c=y_test, cmap=cm_bright,\n edgecolors='k', alpha=0.6)\n\n ax.set_xlim(xx.min(), xx.max())\n ax.set_ylim(yy.min(), yy.max())\n ax.set_xticks(())\n ax.set_yticks(())\n if ds_cnt == 0:\n ax.set_title(name)\n ax.text(xx.max() - .3, yy.min() + .3, ('%.2f' % score).lstrip('0'),\n size=15, horizontalalignment='right')\n i += 1\n\nplt.tight_layout()\nplt.show()", "_____no_output_____" ] ], [ [ "### b) Métricas\n\nDado que el sistema de calcular métricas sigue el mismo formato, solo cambiando el conjunto de datos y el modelo, se decide realizar una clase que automatice este proceso.", "_____no_output_____" ] ], [ [ "from metrics_classification import *\n\nclass SklearnClassificationModels:\n def __init__(self,model,name_model):\n\n self.model = model\n self.name_model = name_model\n \n @staticmethod\n def test_train_model(X,y,n_size):\n X_train, X_test, y_train, y_test = train_test_split(X, y,test_size=n_size , random_state=42)\n return X_train, X_test, y_train, y_test\n \n def fit_model(self,X,y,test_size):\n X_train, X_test, y_train, y_test = self.test_train_model(X,y,test_size )\n return self.model.fit(X_train, y_train) \n \n def df_testig(self,X,y,test_size):\n X_train, X_test, y_train, y_test = self.test_train_model(X,y,test_size )\n model_fit = self.model.fit(X_train, y_train)\n preds = model_fit.predict(X_test)\n df_temp = pd.DataFrame(\n {\n 'y':y_test,\n 'yhat': model_fit.predict(X_test)\n }\n )\n \n return df_temp\n \n def metrics(self,X,y,test_size):\n df_temp = self.df_testig(X,y,test_size)\n df_metrics = summary_metrics(df_temp)\n df_metrics['model'] = self.name_model\n \n return df_metrics\n", "_____no_output_____" ], [ "# metrics \n\nimport itertools\n\n# nombre modelos\nnames_models = [\"Logistic\",\n \"RBF SVM\", \n \"Decision Tree\", \n \"Random Forest\"\n]\n\n# modelos\nclassifiers = [\n LogisticRegression(),\n SVC(gamma=2, C=1),\n DecisionTreeClassifier(max_depth=5),\n RandomForestClassifier(max_depth=5, n_estimators=10, max_features=1),\n]\n\n# datasets\nnames_dataset = ['make_moons',\n 'make_circles',\n 'linearly_separable'\n ]\n\nX, y = make_classification(n_features=2, n_redundant=0, n_informative=2,\n random_state=1, n_clusters_per_class=1)\nrng = np.random.RandomState(2)\nX += 2 * rng.uniform(size=X.shape)\nlinearly_separable = (X, y)\n\ndatasets = [make_moons(noise=0.3, random_state=0),\n make_circles(noise=0.2, factor=0.5, random_state=1),\n linearly_separable\n ]\n\n\n# juntar informacion\nlist_models = list(zip(names_models,classifiers))\nlist_dataset = list(zip(names_dataset,datasets))\n\nframes = []\nfor x in itertools.product(list_models, list_dataset):\n \n name_model = x[0][0]\n classifier = x[0][1]\n \n name_dataset = x[1][0]\n dataset = x[1][1]\n \n X = dataset[0]\n Y = dataset[1]\n \n fit_model = SklearnClassificationModels( classifier,name_model)\n df = fit_model.metrics(X,Y,0.2)\n df['dataset'] = name_dataset\n \n frames.append(df)", "/home/falfaro/.cache/pypoetry/virtualenvs/pymessi-xyyw3p3f-py3.6/lib/python3.6/site-packages/sklearn/metrics/_classification.py:1221: UndefinedMetricWarning: Precision is ill-defined and being set to 0.0 in labels with no predicted samples. Use `zero_division` parameter to control this behavior.\n _warn_prf(average, modifier, msg_start, len(result))\n" ], [ "# juntar resultados\npd.concat(frames)", "_____no_output_____" ] ], [ [ "## Referencia\n\n1. [Supervised learning](https://scikit-learn.org/stable/supervised_learning.html)", "_____no_output_____" ] ] ]

[ "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown" ]

[ [ "markdown", "markdown", "markdown", "markdown", "markdown", "markdown", "markdown", "markdown" ], [ "code", "code", "code" ], [ "markdown" ], [ "code", "code", "code" ], [ "markdown" ], [ "code" ], [ "markdown" ], [ "code" ], [ "markdown", "markdown" ], [ "code", "code", "code", "code", "code" ], [ "markdown", "markdown" ], [ "code" ], [ "markdown" ], [ "code", "code", "code" ], [ "markdown" ] ]

[ "Apache-2.0" ]

[ "Apache-2.0" ]

[ "Apache-2.0" ]

[ [ [ "## Constant features", "_____no_output_____" ] ], [ [ "import numpy as np\nimport pandas as pd\nfrom sklearn.model_selection import train_test_split\nfrom sklearn.feature_selection import VarianceThreshold", "_____no_output_____" ] ], [ [ "## Read Data", "_____no_output_____" ] ], [ [ "data.head(5)", "_____no_output_____" ] ], [ [ "### Train - Test Split", "_____no_output_____" ] ], [ [ "data.convert_dtypes().dtypes", "_____no_output_____" ], [ "# separate dataset into train and test\nX_train, X_test, y_train, y_test = train_test_split(\n data.drop(labels=['Label_code'], axis=1), # drop the target\n data['Label_code'], # just the target\n test_size=0.2,\n random_state=0)\nX_train.shape, X_test.shape", "_____no_output_____" ] ], [ [ "### Using VarianceThreshold from Scikit-learn\n\nThe VarianceThreshold from sklearn provides a simple baseline approach to feature selection. It removes all features which variance doesn’t meet a certain threshold. By default, it removes all zero-variance features, i.e., features that have the same value in all samples.", "_____no_output_____" ] ], [ [ "sel = VarianceThreshold(threshold=0.01)\nsel.fit(X_train) # fit finds the features with zero variance", "_____no_output_____" ], [ "# get_support is a boolean vector that indicates which features are retained\n# if we sum over get_support, we get the number of features that are not constant\n# (if necessary, print the result of sel.get_support() to understand its output)\nsum(sel.get_support())", "_____no_output_____" ], [ "# now let's print the number of constant feautures\n# (see how we use ~ to exclude non-constant features)\nconstant = X_train.columns[~sel.get_support()]\nlen(constant)", "_____no_output_____" ] ], [ [ "We can see that 0 columns / variables are constant. This means that 0 variables show the same value, just one value, for all the observations of the training set.", "_____no_output_____" ] ], [ [ "# let's print the constant variable names\nconstant", "_____no_output_____" ], [ "# let's visualise the values of one of the constant variables\n# as an example\nX_train['Protocol_code'].unique()", "_____no_output_____" ], [ "# we can do the same for every feature:\nfor col in constant:\n print(col, X_train[col].unique())", "_____no_output_____" ] ], [ [ "We then use the transform() method of the VarianceThreshold to reduce the training and testing sets to its non-constant features.\n\nNote that VarianceThreshold returns a NumPy array without feature names, so we need to capture the names first, and reconstitute the dataframe in a later step.", "_____no_output_____" ] ], [ [ "# capture non-constant feature names\nfeat_names = X_train.columns[sel.get_support()]", "_____no_output_____" ], [ "X_train = sel.transform(X_train)\nX_test = sel.transform(X_test)\nX_train.shape, X_test.shape", "_____no_output_____" ] ], [ [ "We have now 23 variables.", "_____no_output_____" ] ], [ [ "# X_ train is a NumPy array\nX_train", "_____no_output_____" ], [ "# reconstitute de dataframe\nX_train = pd.DataFrame(X_train, columns=feat_names)\nX_train.head()", "_____no_output_____" ] ], [ [ "In the Kyoto dataset, 0 features was classified as constant were found, remaining the original 23 features of the dataset", "_____no_output_____" ], [ "## Standardize Data", "_____no_output_____" ] ], [ [ "from sklearn.preprocessing import StandardScaler\nscaler = StandardScaler().fit(X_train)\nX_train = scaler.transform(X_train)", "_____no_output_____" ] ], [ [ "## Hyperparameter Optimization", "_____no_output_____" ] ], [ [ "import numpy as np\nimport pandas as pd\nfrom sklearn.model_selection import GridSearchCV\n\nclass EstimatorSelectionHelper:\n \n def __init__(self, models, params):\n self.models = models\n self.params = params\n self.keys = models.keys()\n self.grid_searches = {}\n \n def fit(self, X, y, **grid_kwargs):\n for key in self.keys:\n print('Running GridSearchCV for %s.' % key)\n model = self.models[key]\n params = self.params[key]\n grid_search = GridSearchCV(model, params, **grid_kwargs)\n grid_search.fit(X, y)\n self.grid_searches[key] = grid_search\n print('Done.')\n \n def score_summary(self, sort_by='mean_test_score'):\n frames = []\n for name, grid_search in self.grid_searches.items():\n frame = pd.DataFrame(grid_search.cv_results_)\n frame = frame.filter(regex='^(?!.*param_).*$')\n frame['estimator'] = len(frame)*[name]\n frames.append(frame)\n df = pd.concat(frames)\n \n df = df.sort_values([sort_by], ascending=False)\n df = df.reset_index()\n df = df.drop(['rank_test_score', 'index'], 1)\n \n columns = df.columns.tolist()\n columns.remove('estimator')\n columns = ['estimator']+columns\n df = df[columns]\n return df", "_____no_output_____" ], [ "from sklearn import linear_model\nfrom sklearn.naive_bayes import GaussianNB\nfrom sklearn.ensemble import RandomForestClassifier\nfrom sklearn.neighbors import KNeighborsClassifier\nfrom catboost import CatBoostClassifier\n\n\nmodels = { \n 'LogisticRegression': linear_model.LogisticRegression(max_iter=1000),\n 'GaussianNB': GaussianNB(),\n 'RandomForest': RandomForestClassifier(random_state=123),\n 'KNN': KNeighborsClassifier(n_jobs=-1),\n 'CatBoost': CatBoostClassifier(),\n}\n\nparams = { \n 'LogisticRegression': {'C':[0.1,0.5,1,2,3,4,5,10,20,25]},\n 'GaussianNB': {'var_smoothing':[1e-9,1e-8,1e-7,1e-6,1e-5,1e-4]},\n 'RandomForest': {'max_depth': [70,80,90,100],'n_estimators': [100,1000]},\n 'KNN': {'n_neighbors':[2,3,4,5],'leaf_size':[1,2,3],'weights':['uniform', 'distance'],\n 'algorithm':['auto','ball_tree','kd_tree','brute']},\n 'CatBoost': {'depth':[4,5,6,7],'learning_rate':[0.01,0.02,0.03,0.04],'iterations':[10,20,30,40,50]}\n}", "_____no_output_____" ], [ "%%time\nhelper = EstimatorSelectionHelper(models, params)\nhelper.fit(X_test, y_test, scoring='f1', n_jobs=2)", "Running GridSearchCV for LogisticRegression.\n" ], [ "helper.score_summary()", "_____no_output_____" ], [ "df_gridsearchcv_summary = helper.score_summary()", "_____no_output_____" ], [ "df_gridsearchcv_summary.to_csv(\"gridsearchcv_summaryBase.csv\", index=False)", "_____no_output_____" ] ], [ [ "## Classifiers", "_____no_output_____" ] ], [ [ "from sklearn import linear_model\nfrom sklearn.naive_bayes import GaussianNB\nfrom sklearn.ensemble import RandomForestClassifier\nfrom sklearn.neighbors import KNeighborsClassifier\nfrom catboost import CatBoostClassifier", "_____no_output_____" ] ], [ [ "## Metrics Evaluation", "_____no_output_____" ] ], [ [ "from sklearn.metrics import accuracy_score\nfrom sklearn.metrics import confusion_matrix\nfrom sklearn.metrics import roc_curve, f1_score\nfrom sklearn import metrics\nfrom sklearn.model_selection import cross_val_score", "_____no_output_____" ] ], [ [ "### Logistic Regression", "_____no_output_____" ] ], [ [ "%%time\nclf_LR = linear_model.LogisticRegression(n_jobs=-1, random_state=42, C=1).fit(X_train, y_train)", "CPU times: user 77.6 ms, sys: 211 ms, total: 289 ms\nWall time: 3.2 s\n" ], [ "pred_y_test = clf_LR.predict(X_test)\nprint('Accuracy:', accuracy_score(y_test, pred_y_test))\n\nf1 = f1_score(y_test, pred_y_test)\nprint('F1 Score:', f1)\n\nfpr, tpr, thresholds = roc_curve(y_test, pred_y_test)\nprint('FPR:', fpr[1])\nprint('TPR:', tpr[1])", "Accuracy: 0.4527830397807424\nF1 Score: 0.2463502636691646\nFPR: 0.5989054991991457\nTPR: 0.9503211991434689\n" ] ], [ [ "### Naive Bayes", "_____no_output_____" ] ], [ [ "%%time\nclf_NB = GaussianNB(var_smoothing=1e-05).fit(X_train, y_train)", "CPU times: user 40.1 ms, sys: 7.54 ms, total: 47.7 ms\nWall time: 45.9 ms\n" ], [ "pred_y_testNB = clf_NB.predict(X_test)\nprint('Accuracy:', accuracy_score(y_test, pred_y_testNB))\n\nf1 = f1_score(y_test, pred_y_testNB)\nprint('F1 Score:', f1)\n\nfpr, tpr, thresholds = roc_curve(y_test, pred_y_testNB)\nprint('FPR:', fpr[1])\nprint('TPR:', tpr[1])", "Accuracy: 0.9045584619725122\nF1 Score: 0.0\nFPR: 0.0014682327816337426\nTPR: 0.0\n" ] ], [ [ "### Random Forest", "_____no_output_____" ] ], [ [ "%%time\nclf_RF = RandomForestClassifier(random_state=0,max_depth=70,n_estimators=100).fit(X_train, y_train)", "CPU times: user 6.16 s, sys: 45.8 ms, total: 6.2 s\nWall time: 6.2 s\n" ], [ "pred_y_testRF = clf_RF.predict(X_test)\nprint('Accuracy:', accuracy_score(y_test, pred_y_testRF))\n\nf1 = f1_score(y_test, pred_y_testRF, average='weighted', zero_division=0)\nprint('F1 Score:', f1)\n\nfpr, tpr, thresholds = roc_curve(y_test, pred_y_testRF)\nprint('FPR:', fpr[1])\nprint('TPR:', tpr[1])", "Accuracy: 0.9058885171899561\nF1 Score: 0.8611563563923045\nFPR: 1.0\nTPR: 1.0\n" ] ], [ [ "### KNN", "_____no_output_____" ] ], [ [ "%%time\nclf_KNN = KNeighborsClassifier(algorithm='auto',leaf_size=1,n_neighbors=2,weights='uniform').fit(X_train, y_train)", "CPU times: user 9.96 s, sys: 61.3 ms, total: 10 s\nWall time: 9.98 s\n" ], [ "pred_y_testKNN = clf_KNN.predict(X_test)\nprint('accuracy_score:', accuracy_score(y_test, pred_y_testKNN))\n\nf1 = f1_score(y_test, pred_y_testKNN)\nprint('f1:', f1)\n\nfpr, tpr, thresholds = roc_curve(y_test, pred_y_testKNN)\nprint('fpr:', fpr[1])\nprint('tpr:', tpr[1])", "accuracy_score: 0.9058885171899561\nf1: 0.0\nfpr: 1.0\ntpr: 1.0\n" ] ], [ [ "### CatBoost", "_____no_output_____" ] ], [ [ "%%time\nclf_CB = CatBoostClassifier(random_state=0,depth=7,iterations=50,learning_rate=0.04).fit(X_train, y_train)", "0:\tlearn: 0.5950021\ttotal: 21.5ms\tremaining: 1.05s\n1:\tlearn: 0.4943009\ttotal: 39.9ms\tremaining: 958ms\n2:\tlearn: 0.4235946\ttotal: 57.5ms\tremaining: 901ms\n3:\tlearn: 0.3620619\ttotal: 75.7ms\tremaining: 871ms\n4:\tlearn: 0.3121757\ttotal: 93.6ms\tremaining: 842ms\n5:\tlearn: 0.2371811\ttotal: 111ms\tremaining: 816ms\n6:\tlearn: 0.2070297\ttotal: 129ms\tremaining: 795ms\n7:\tlearn: 0.1610239\ttotal: 148ms\tremaining: 779ms\n8:\tlearn: 0.1441127\ttotal: 166ms\tremaining: 756ms\n9:\tlearn: 0.1140228\ttotal: 186ms\tremaining: 744ms\n10:\tlearn: 0.1028733\ttotal: 203ms\tremaining: 721ms\n11:\tlearn: 0.0790135\ttotal: 221ms\tremaining: 701ms\n12:\tlearn: 0.0614288\ttotal: 239ms\tremaining: 681ms\n13:\tlearn: 0.0478516\ttotal: 259ms\tremaining: 666ms\n14:\tlearn: 0.0375712\ttotal: 277ms\tremaining: 647ms\n15:\tlearn: 0.0312555\ttotal: 295ms\tremaining: 627ms\n16:\tlearn: 0.0249900\ttotal: 312ms\tremaining: 606ms\n17:\tlearn: 0.0196449\ttotal: 330ms\tremaining: 586ms\n18:\tlearn: 0.0159795\ttotal: 348ms\tremaining: 567ms\n19:\tlearn: 0.0128696\ttotal: 365ms\tremaining: 548ms\n20:\tlearn: 0.0106125\ttotal: 384ms\tremaining: 530ms\n21:\tlearn: 0.0090042\ttotal: 402ms\tremaining: 512ms\n22:\tlearn: 0.0074177\ttotal: 420ms\tremaining: 493ms\n23:\tlearn: 0.0062532\ttotal: 438ms\tremaining: 474ms\n24:\tlearn: 0.0052067\ttotal: 456ms\tremaining: 456ms\n25:\tlearn: 0.0044424\ttotal: 474ms\tremaining: 437ms\n26:\tlearn: 0.0037750\ttotal: 491ms\tremaining: 418ms\n27:\tlearn: 0.0032876\ttotal: 508ms\tremaining: 399ms\n28:\tlearn: 0.0028467\ttotal: 526ms\tremaining: 381ms\n29:\tlearn: 0.0024856\ttotal: 543ms\tremaining: 362ms\n30:\tlearn: 0.0022038\ttotal: 561ms\tremaining: 344ms\n31:\tlearn: 0.0019416\ttotal: 580ms\tremaining: 326ms\n32:\tlearn: 0.0017309\ttotal: 598ms\tremaining: 308ms\n33:\tlearn: 0.0015386\ttotal: 615ms\tremaining: 289ms\n34:\tlearn: 0.0013602\ttotal: 632ms\tremaining: 271ms\n35:\tlearn: 0.0012191\ttotal: 650ms\tremaining: 253ms\n36:\tlearn: 0.0010849\ttotal: 668ms\tremaining: 235ms\n37:\tlearn: 0.0009851\ttotal: 687ms\tremaining: 217ms\n38:\tlearn: 0.0008939\ttotal: 704ms\tremaining: 199ms\n39:\tlearn: 0.0008057\ttotal: 722ms\tremaining: 180ms\n40:\tlearn: 0.0007422\ttotal: 739ms\tremaining: 162ms\n41:\tlearn: 0.0006801\ttotal: 756ms\tremaining: 144ms\n42:\tlearn: 0.0006254\ttotal: 774ms\tremaining: 126ms\n43:\tlearn: 0.0005768\ttotal: 791ms\tremaining: 108ms\n44:\tlearn: 0.0005253\ttotal: 808ms\tremaining: 89.8ms\n45:\tlearn: 0.0004894\ttotal: 827ms\tremaining: 71.9ms\n46:\tlearn: 0.0004579\ttotal: 846ms\tremaining: 54ms\n47:\tlearn: 0.0004291\ttotal: 865ms\tremaining: 36ms\n48:\tlearn: 0.0004010\ttotal: 883ms\tremaining: 18ms\n49:\tlearn: 0.0003927\ttotal: 899ms\tremaining: 0us\nCPU times: user 6.4 s, sys: 1.79 s, total: 8.18 s\nWall time: 961 ms\n" ], [ "pred_y_testCB = clf_CB.predict(X_test)\nprint('Accuracy:', accuracy_score(y_test, pred_y_testCB))\n\nf1 = f1_score(y_test, pred_y_testCB, average='weighted', zero_division=0)\nprint('F1 Score:', f1)\n\nfpr, tpr, thresholds = roc_curve(y_test, pred_y_testCB)\nprint('FPR:', fpr[1])\nprint('TPR:', tpr[1])", "Accuracy: 0.9058885171899561\nF1 Score: 0.8611563563923045\nFPR: 1.0\nTPR: 1.0\n" ] ], [ [ "## Model Evaluation", "_____no_output_____" ] ], [ [ "import pandas as pd, numpy as np\ntest_df = pd.read_csv(\"../Kyoto_Test.csv\")\ntest_df.shape", "_____no_output_____" ], [ "# Create feature matrix X and target vextor y\ny_eval = test_df['Label_code']\nX_eval = test_df.drop(columns=['Label_code'])", "_____no_output_____" ] ], [ [ "### Model Evaluation - Logistic Regression", "_____no_output_____" ] ], [ [ "modelLR = linear_model.LogisticRegression(n_jobs=-1, random_state=42, C=1)\nmodelLR.fit(X_train, y_train)", "_____no_output_____" ], [ "# Predict on the new unseen test data\ny_evalpredLR = modelLR.predict(X_eval)\ny_predLR = modelLR.predict(X_test)", "_____no_output_____" ], [ "train_scoreLR = modelLR.score(X_train, y_train)\ntest_scoreLR = modelLR.score(X_test, y_test)\nprint(\"Training accuracy is \", train_scoreLR)\nprint(\"Testing accuracy is \", test_scoreLR)", "Training accuracy is 0.9304038531296602\nTesting accuracy is 0.4527830397807424\n" ], [ "from sklearn.metrics import confusion_matrix, precision_score, recall_score, f1_score\nprint('Performance measures for test:')\nprint('--------')\nprint('Accuracy:', test_scoreLR)\nprint('F1 Score:',f1_score(y_test, y_predLR))\nprint('Precision Score:',precision_score(y_test, y_predLR))\nprint('Recall Score:', recall_score(y_test, y_predLR))\nprint('Confusion Matrix:\\n', confusion_matrix(y_test, y_predLR))", "Performance measures for test:\n--------\nAccuracy: 0.4527830397807424\nF1 Score: 0.2463502636691646\nPrecision Score: 0.14151785714285714\nRecall Score: 0.9503211991434689\nConfusion Matrix:\n [[ 9015 13461]\n [ 116 2219]]\n" ] ], [ [ "### Cross validation - Logistic Regression", "_____no_output_____" ] ], [ [ "from sklearn.model_selection import cross_val_score\nfrom sklearn import metrics\n\naccuracy = cross_val_score(modelLR, X_eval, y_eval, cv=10, scoring='accuracy')\nprint(\"Accuracy: %0.5f (+/- %0.5f)\" % (accuracy.mean(), accuracy.std() * 2))\n\nf = cross_val_score(modelLR, X_eval, y_eval, cv=10, scoring='f1')\nprint(\"F1 Score: %0.5f (+/- %0.5f)\" % (f.mean(), f.std() * 2))\n\nprecision = cross_val_score(modelLR, X_eval, y_eval, cv=10, scoring='precision')\nprint(\"Precision: %0.5f (+/- %0.5f)\" % (precision.mean(), precision.std() * 2))\n\nrecall = cross_val_score(modelLR, X_eval, y_eval, cv=10, scoring='recall')\nprint(\"Recall: %0.5f (+/- %0.5f)\" % (recall.mean(), recall.std() * 2))", "Accuracy: 0.90021 (+/- 0.00142)\nF1 Score: 0.00064 (+/- 0.00385)\nPrecision: 0.00769 (+/- 0.04615)\nRecall: 0.00033 (+/- 0.00201)\n" ] ], [ [ "### Model Evaluation - Naive Bayes", "_____no_output_____" ] ], [ [ "modelNB = GaussianNB(var_smoothing=1e-05)\nmodelNB.fit(X_train, y_train)", "_____no_output_____" ], [ "# Predict on the new unseen test data\ny_evalpredNB = modelNB.predict(X_eval)\ny_predNB = modelNB.predict(X_test)", "_____no_output_____" ], [ "train_scoreNB = modelNB.score(X_train, y_train)\ntest_scoreNB = modelNB.score(X_test, y_test)\nprint(\"Training accuracy is \", train_scoreNB)\nprint(\"Testing accuracy is \", test_scoreNB)", "Training accuracy is 0.3204626979968562\nTesting accuracy is 0.9045584619725122\n" ], [ "from sklearn.metrics import confusion_matrix, precision_score, recall_score, f1_score\nprint('Performance measures for test:')\nprint('--------')\nprint('Accuracy:', test_scoreNB)\nprint('F1 Score:',f1_score(y_test, y_predNB))\nprint('Precision Score:',precision_score(y_test, y_predNB))\nprint('Recall Score:', recall_score(y_test, y_predNB))\nprint('Confusion Matrix:\\n', confusion_matrix(y_test, y_predNB))", "Performance measures for test:\n--------\nAccuracy: 0.9045584619725122\nF1 Score: 0.0\nPrecision Score: 0.0\nRecall Score: 0.0\nConfusion Matrix:\n [[22443 33]\n [ 2335 0]]\n" ] ], [ [ "### Cross validation - Naive Bayes", "_____no_output_____" ] ], [ [ "from sklearn.model_selection import cross_val_score\nfrom sklearn import metrics\n\naccuracy = cross_val_score(modelNB, X_eval, y_eval, cv=10, scoring='accuracy')\nprint(\"Accuracy: %0.5f (+/- %0.5f)\" % (accuracy.mean(), accuracy.std() * 2))\n\nf = cross_val_score(modelNB, X_eval, y_eval, cv=10, scoring='f1')\nprint(\"F1 Score: %0.5f (+/- %0.5f)\" % (f.mean(), f.std() * 2))\n\nprecision = cross_val_score(modelNB, X_eval, y_eval, cv=10, scoring='precision')\nprint(\"Precision: %0.5f (+/- %0.5f)\" % (precision.mean(), precision.std() * 2))\n\nrecall = cross_val_score(modelNB, X_eval, y_eval, cv=10, scoring='recall')\nprint(\"Recall: %0.5f (+/- %0.5f)\" % (recall.mean(), recall.std() * 2))", "Accuracy: 0.51851 (+/- 0.28039)\nF1 Score: 0.25979 (+/- 0.02808)\nPrecision: 0.21404 (+/- 0.38891)\nRecall: 0.86306 (+/- 0.46302)\n" ] ], [ [ "### Model Evaluation - Random Forest", "_____no_output_____" ] ], [ [ "modelRF = RandomForestClassifier(random_state=0,max_depth=70,n_estimators=100)\nmodelRF.fit(X_train, y_train)", "_____no_output_____" ], [ "# Predict on the new unseen test data\ny_evalpredRF = modelRF.predict(X_eval)\ny_predRF = modelRF.predict(X_test)", "_____no_output_____" ], [ "train_scoreRF = modelRF.score(X_train, y_train)\ntest_scoreRF = modelRF.score(X_test, y_test)\nprint(\"Training accuracy is \", train_scoreRF)\nprint(\"Testing accuracy is \", test_scoreRF)", "Training accuracy is 1.0\nTesting accuracy is 0.9058885171899561\n" ], [ "import warnings\nwarnings.filterwarnings(\"ignore\", category=DeprecationWarning)", "_____no_output_____" ], [ "from sklearn.metrics import confusion_matrix, precision_score, recall_score, f1_score\nprint('Performance measures for test:')\nprint('--------')\nprint('Accuracy:', test_scoreRF)\nprint('F1 Score:', f1_score(y_test, y_predRF, average='weighted', zero_division=1))\nprint('Precision Score:', precision_score(y_test, y_predRF, average='weighted', zero_division=1))\nprint('Recall Score:', recall_score(y_test, y_predRF, average='weighted', zero_division=1))\nprint('Confusion Matrix:\\n', confusion_matrix(y_test, y_predRF))", "Performance measures for test:\n--------\nAccuracy: 0.9058885171899561\nF1 Score: 0.8611563563923045\nPrecision Score: 0.9147454883866614\nRecall Score: 0.9058885171899561\nConfusion Matrix:\n [[22476 0]\n [ 2335 0]]\n" ] ], [ [ "### Cross validation - Random Forest", "_____no_output_____" ] ], [ [ "from sklearn.model_selection import cross_val_score\nfrom sklearn import metrics\n\naccuracy = cross_val_score(modelRF, X_eval, y_eval, cv=10, scoring='accuracy')\nprint(\"Accuracy: %0.5f (+/- %0.5f)\" % (accuracy.mean(), accuracy.std() * 2))\n\nf = cross_val_score(modelRF, X_eval, y_eval, cv=10, scoring='f1')\nprint(\"F1 Score: %0.5f (+/- %0.5f)\" % (f.mean(), f.std() * 2))\n\nprecision = cross_val_score(modelRF, X_eval, y_eval, cv=10, scoring='precision')\nprint(\"Precision: %0.5f (+/- %0.5f)\" % (precision.mean(), precision.std() * 2))\n\nrecall = cross_val_score(modelRF, X_eval, y_eval, cv=10, scoring='recall')\nprint(\"Recall: %0.5f (+/- %0.5f)\" % (recall.mean(), recall.std() * 2))", "Accuracy: 0.99929 (+/- 0.00060)\nF1 Score: 0.99631 (+/- 0.00312)\nPrecision: 0.99950 (+/- 0.00154)\nRecall: 0.99315 (+/- 0.00676)\n" ] ], [ [ "### Model Evaluation - KNN", "_____no_output_____" ] ], [ [ "modelKNN = KNeighborsClassifier(algorithm='auto',leaf_size=1,n_neighbors=2,weights='uniform')\nmodelKNN.fit(X_train, y_train)", "_____no_output_____" ], [ "# Predict on the new unseen test data\ny_evalpredKNN = modelKNN.predict(X_eval)\ny_predKNN = modelKNN.predict(X_test)", "_____no_output_____" ], [ "train_scoreKNN = modelKNN.score(X_train, y_train)\ntest_scoreKNN = modelKNN.score(X_test, y_test)\nprint(\"Training accuracy is \", train_scoreKNN)\nprint(\"Testing accuracy is \", test_scoreKNN)", "_____no_output_____" ], [ "from sklearn.metrics import confusion_matrix, precision_score, recall_score, f1_score\nprint('Performance measures for test:')\nprint('--------')\nprint('Accuracy:', test_scoreKNN)\nprint('F1 Score:', f1_score(y_test, y_predKNN, average='weighted', zero_division=1))\nprint('Precision Score:', precision_score(y_test, y_predKNN, average='weighted', zero_division=1))\nprint('Recall Score:', recall_score(y_test, y_predKNN, average='weighted', zero_division=1))\nprint('Confusion Matrix:\\n', confusion_matrix(y_test, y_predKNN))", "_____no_output_____" ] ], [ [ "### Cross validation - KNN", "_____no_output_____" ] ], [ [ "from sklearn.model_selection import cross_val_score\nfrom sklearn import metrics\n\naccuracy = cross_val_score(modelKNN, X_eval, y_eval, cv=10, scoring='accuracy')\nprint(\"Accuracy: %0.5f (+/- %0.5f)\" % (accuracy.mean(), accuracy.std() * 2))\n\nf = cross_val_score(modelKNN, X_eval, y_eval, cv=10, scoring='f1')\nprint(\"F1 Score: %0.5f (+/- %0.5f)\" % (f.mean(), f.std() * 2))\n\nprecision = cross_val_score(modelKNN, X_eval, y_eval, cv=10, scoring='precision')\nprint(\"Precision: %0.5f (+/- %0.5f)\" % (precision.mean(), precision.std() * 2))\n\nrecall = cross_val_score(modelKNN, X_eval, y_eval, cv=10, scoring='recall')\nprint(\"Recall: %0.5f (+/- %0.5f)\" % (recall.mean(), recall.std() * 2))", "_____no_output_____" ] ], [ [ "### Model Evaluation - CatBoost", "_____no_output_____" ] ], [ [ "modelCB = CatBoostClassifier(random_state=0,depth=7,iterations=50,learning_rate=0.04)\nmodelCB.fit(X_train, y_train)", "_____no_output_____" ], [ "# Predict on the new unseen test data\ny_evalpredCB = modelCB.predict(X_eval)\ny_predCB = modelCB.predict(X_test)", "_____no_output_____" ], [ "train_scoreCB = modelCB.score(X_train, y_train)\ntest_scoreCB = modelCB.score(X_test, y_test)\nprint(\"Training accuracy is \", train_scoreCB)\nprint(\"Testing accuracy is \", test_scoreCB)", "_____no_output_____" ], [ "from sklearn.metrics import confusion_matrix, precision_score, recall_score, f1_score\nprint('Performance measures for test:')\nprint('--------')\nprint('Accuracy:', test_scoreCB)\nprint('F1 Score:',f1_score(y_test, y_predCB, average='weighted', zero_division=1))\nprint('Precision Score:',precision_score(y_test, y_predCB, average='weighted', zero_division=1))\nprint('Recall Score:', recall_score(y_test, y_predCB, average='weighted', zero_division=1))\nprint('Confusion Matrix:\\n', confusion_matrix(y_test, y_predCB))", "_____no_output_____" ] ], [ [ "### Cross validation - CatBoost", "_____no_output_____" ] ], [ [ "from sklearn.model_selection import cross_val_score\nfrom sklearn import metrics\n\naccuracy = cross_val_score(modelCB, X_eval, y_eval, cv=10, scoring='accuracy')\nf = cross_val_score(modelCB, X_eval, y_eval, cv=10, scoring='f1')\nprecision = cross_val_score(modelCB, X_eval, y_eval, cv=10, scoring='precision')\nrecall = cross_val_score(modelCB, X_eval, y_eval, cv=10, scoring='recall')", "_____no_output_____" ], [ "print(\"Accuracy: %0.5f (+/- %0.5f)\" % (accuracy.mean(), accuracy.std() * 2))\nprint(\"F1 Score: %0.5f (+/- %0.5f)\" % (f.mean(), f.std() * 2))\nprint(\"Precision: %0.5f (+/- %0.5f)\" % (precision.mean(), precision.std() * 2))\nprint(\"Recall: %0.5f (+/- %0.5f)\" % (recall.mean(), recall.std() * 2))", "_____no_output_____" ] ] ]

[ "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code" ]

[ [ "markdown" ], [ "code" ], [ "markdown" ], [ "code" ], [ "markdown" ], [ "code", "code" ], [ "markdown" ], [ "code", "code", "code" ], [ "markdown" ], [ "code", "code", "code" ], [ "markdown" ], [ "code", "code" ], [ "markdown" ], [ "code", "code" ], [ "markdown", "markdown" ], [ "code" ], [ "markdown" ], [ "code", "code", "code", "code", "code", "code" ], [ "markdown" ], [ "code" ], [ "markdown" ], [ "code" ], [ "markdown" ], [ "code", "code" ], [ "markdown" ], [ "code", "code" ], [ "markdown" ], [ "code", "code" ], [ "markdown" ], [ "code", "code" ], [ "markdown" ], [ "code", "code" ], [ "markdown" ], [ "code", "code" ], [ "markdown" ], [ "code", "code", "code", "code" ], [ "markdown" ], [ "code" ], [ "markdown" ], [ "code", "code", "code", "code" ], [ "markdown" ], [ "code" ], [ "markdown" ], [ "code", "code", "code", "code", "code" ], [ "markdown" ], [ "code" ], [ "markdown" ], [ "code", "code", "code", "code" ], [ "markdown" ], [ "code" ], [ "markdown" ], [ "code", "code", "code", "code" ], [ "markdown" ], [ "code", "code" ] ]

[ "MIT" ]

[ "MIT" ]

[ "MIT" ]

[ [ [ "amazon_rekognition_labels = ['Angry',\n 'Angry',\n 'Fear',\n 'Disgust',\n 'Sad',\n 'Neutral',\n 'Surprise',\n 'Angry',\n 'Fear',\n 'Angry',\n 'Angry',\n 'Angry',\n 'Angry',\n 'Angry',\n 'Angry',\n 'Angry',\n 'Angry',\n 'Angry',\n 'Angry',\n 'Angry',\n 'Surprise',\n 'Neutral',\n 'Neutral',\n 'Neutral',\n 'Fear',\n 'Neutral',\n 'Neutral',\n 'Angry',\n 'Angry',\n 'Neutral',\n 'Sad',\n 'Fear',\n 'Sad',\n 'Disgust',\n 'Disgust',\n 'Neutral',\n 'Angry',\n 'Angry',\n 'Angry',\n 'Angry',\n 'Angry',\n 'Neutral',\n 'Angry',\n 'Neutral',\n 'Neutral',\n 'Angry',\n 'Fear',\n 'Angry',\n 'Angry',\n 'Angry',\n 'Disgust',\n 'Angry',\n 'Angry',\n 'Angry',\n 'Neutral',\n 'Fear',\n 'Neutral',\n 'Angry',\n 'Angry',\n 'Angry',\n 'Sad',\n 'Neutral',\n 'Angry',\n 'Angry',\n 'Angry',\n 'Angry',\n 'Angry',\n 'Neutral',\n 'Angry',\n 'Surprise',\n 'Angry',\n 'Angry',\n 'Angry',\n 'Angry',\n 'Angry',\n 'Neutral',\n 'Angry',\n 'Disgust',\n 'Angry',\n 'Angry',\n 'Neutral',\n 'Neutral',\n 'Sad',\n 'Angry',\n 'Angry',\n 'Neutral',\n 'Sad',\n 'Neutral',\n 'Neutral',\n 'Neutral',\n 'Sad',\n 'Neutral',\n 'Disgust',\n 'Angry',\n 'Angry',\n 'Angry',\n 'Angry',\n 'Angry',\n 'Angry',\n 'Neutral',\n 'Surprise',\n 'Disgust',\n 'Angry',\n 'Angry',\n 'Neutral',\n 'Disgust',\n 'Fear',\n 'Disgust',\n 'Disgust',\n 'Angry',\n 'Neutral',\n 'Disgust',\n 'Disgust',\n 'Disgust',\n 'Angry',\n 'Neutral',\n 'Disgust',\n 'Sad',\n 'Angry',\n 'Disgust',\n 'Angry',\n 'Disgust',\n 'Disgust',\n 'Angry',\n 'Neutral',\n 'Angry',\n 'Neutral',\n 'Disgust',\n 'Disgust',\n 'Disgust',\n 'Disgust',\n 'Neutral',\n 'Neutral',\n 'Disgust',\n 'Disgust',\n 'Angry',\n 'Disgust',\n 'Angry',\n 'Neutral',\n 'Neutral',\n 'Neutral',\n 'Disgust',\n 'Disgust',\n 'Disgust',\n 'Fear',\n 'Surprise',\n 'Disgust',\n 'Disgust',\n 'Disgust',\n 'Disgust',\n 'Neutral',\n 'Disgust',\n 'Angry',\n 'Neutral',\n 'Fear',\n 'Angry',\n 'Disgust',\n 'Disgust',\n 'Disgust',\n 'Disgust',\n 'Angry',\n 'Disgust',\n 'Angry',\n 'Disgust',\n 'Angry',\n 'Surprise',\n 'Disgust',\n 'Angry',\n 'Disgust',\n 'Neutral',\n 'Disgust',\n 'Neutral',\n 'Sad',\n 'Angry',\n 'Fear',\n 'Surprise',\n 'Angry',\n 'Angry',\n 'Disgust',\n 'Disgust',\n 'Disgust',\n 'Neutral',\n 'Angry',\n 'Neutral',\n 'Neutral',\n 'Neutral',\n 'Neutral',\n 'Disgust',\n 'Disgust',\n 'Disgust',\n 'Disgust',\n 'Disgust',\n 'Neutral',\n 'Disgust',\n 'Disgust',\n 'Disgust',\n 'Disgust',\n 'Happy',\n 'Neutral',\n 'Angry',\n 'Angry',\n 'Fear',\n 'Fear',\n 'Neutral',\n 'Surprise',\n 'Angry',\n 'Angry',\n 'Fear',\n 'Surprise',\n 'Fear',\n 'Fear',\n 'Neutral',\n 'Fear',\n 'Fear',\n 'Neutral',\n 'Fear',\n 'Neutral',\n 'Surprise',\n 'Sad',\n 'Fear',\n 'Angry',\n 'Neutral',\n 'Fear',\n 'Neutral',\n 'Sad',\n 'Fear',\n 'Surprise',\n 'Disgust',\n 'Sad',\n 'Neutral',\n 'Sad',\n 'Surprise',\n 'Sad',\n 'Sad',\n 'Sad',\n 'Fear',\n 'Fear',\n 'Happy',\n 'Sad',\n 'Sad',\n 'Fear',\n 'Sad',\n 'Angry',\n 'Fear',\n 'Sad',\n 'Neutral',\n 'Fear',\n 'Sad',\n 'Sad',\n 'Sad',\n 'Angry',\n 'Fear',\n 'Neutral',\n 'Sad',\n 'Surprise',\n 'Neutral',\n 'Sad',\n 'Fear',\n 'Fear',\n 'Neutral',\n 'Fear',\n 'Angry',\n 'Sad',\n 'Fear',\n 'Angry',\n 'Surprise',\n 'Fear',\n 'Sad',\n 'Neutral',\n 'Neutral',\n 'Surprise',\n 'Angry',\n 'Sad',\n 'Happy',\n 'Angry',\n 'Fear',\n 'Neutral',\n 'Angry',\n 'Disgust',\n 'Angry',\n 'Neutral',\n 'Surprise',\n 'Fear',\n 'Sad',\n 'Fear',\n 'Sad',\n 'Neutral',\n 'Neutral',\n 'Surprise',\n 'Sad',\n 'Neutral',\n 'Surprise',\n 'Fear',\n 'Fear',\n 'Neutral',\n 'Surprise',\n 'Neutral',\n 'Surprise',\n 'Sad',\n 'Surprise',\n 'Happy',\n 'Neutral',\n 'Surprise',\n 'Happy',\n 'Neutral',\n 'Happy',\n 'Happy',\n 'Disgust',\n 'Happy',\n 'Surprise',\n 'Happy',\n 'Happy',\n 'Happy',\n 'Surprise',\n 'Happy',\n 'Surprise',\n 'Neutral',\n 'Happy',\n 'Happy',\n 'Happy',\n 'Happy',\n 'Happy',\n 'Happy',\n 'Fear',\n 'Happy',\n 'Happy',\n 'Happy',\n 'Happy',\n 'Happy',\n 'Happy',\n 'Happy',\n 'Happy',\n 'Happy',\n 'Happy',\n 'Angry',\n 'Happy',\n 'Angry',\n 'Neutral',\n 'Happy',\n 'Disgust',\n 'Happy',\n 'Happy',\n 'Happy',\n 'Happy',\n 'Happy',\n 'Surprise',\n 'Surprise',\n 'Happy',\n 'Happy',\n 'Sad',\n 'Happy',\n 'Happy',\n 'Happy',\n 'Disgust',\n 'Happy',\n 'Happy',\n 'Happy',\n 'Happy',\n 'Surprise',\n 'Happy',\n 'Happy',\n 'Happy',\n 'Surprise',\n 'Surprise',\n 'Happy',\n 'Happy',\n 'Happy',\n 'Happy',\n 'Happy',\n 'Happy',\n 'Happy',\n 'Surprise',\n 'Happy',\n 'Happy',\n 'Happy',\n 'Disgust',\n 'Surprise',\n 'Happy',\n 'Happy',\n 'Happy',\n 'Surprise',\n 'Disgust',\n 'Happy',\n 'Happy',\n 'Happy',\n 'Happy',\n 'Surprise',\n 'Surprise',\n 'Happy',\n 'Disgust',\n 'Happy',\n 'Happy',\n 'Happy',\n 'Happy',\n 'Happy',\n 'Happy',\n 'Happy',\n 'Happy',\n 'Surprise',\n 'Happy',\n 'Neutral',\n 'Sad',\n 'Sad',\n 'Neutral',\n 'Sad',\n 'Angry',\n 'Sad',\n 'Neutral',\n 'Neutral',\n 'Sad',\n 'Sad',\n 'Neutral',\n 'Angry',\n 'Neutral',\n 'Neutral',\n 'Disgust',\n 'Neutral',\n 'Sad',\n 'Sad',\n 'Angry',\n 'Surprise',\n 'Sad',\n 'Neutral',\n 'Sad',\n 'Neutral',\n 'Neutral',\n 'Neutral',\n 'Disgust',\n 'Neutral',\n 'Disgust',\n 'Neutral',\n 'Surprise',\n 'Sad',\n 'Neutral',\n 'Neutral',\n 'Sad',\n 'Neutral',\n 'Sad',\n 'Neutral',\n 'Neutral',\n 'Neutral',\n 'Surprise',\n 'Fear',\n 'Angry',\n 'Sad',\n 'Neutral',\n 'Neutral',\n 'Angry',\n 'Neutral',\n 'Neutral',\n 'Fear',\n 'Neutral',\n 'Neutral',\n 'Neutral',\n 'Angry',\n 'Sad',\n 'Sad',\n 'Sad',\n 'Angry',\n 'Sad',\n 'Neutral',\n 'Neutral',\n 'Neutral',\n 'Sad',\n 'Neutral',\n 'Disgust',\n 'Neutral',\n 'Neutral',\n 'Sad',\n 'Sad',\n 'Neutral',\n 'Neutral',\n 'Sad',\n 'Sad',\n 'Neutral',\n 'Neutral',\n 'Neutral',\n 'Neutral',\n 'Sad',\n 'Neutral',\n 'Sad',\n 'Sad',\n 'Sad',\n 'Sad',\n 'Neutral',\n 'Sad',\n 'Neutral',\n 'Neutral',\n 'Neutral',\n 'Neutral',\n 'Angry',\n 'Fear',\n 'Sad',\n 'Sad',\n 'Neutral',\n 'Sad',\n 'Happy',\n 'Disgust',\n 'Sad',\n 'Neutral',\n 'Fear',\n 'Surprise',\n 'Surprise',\n 'Surprise',\n 'Surprise',\n 'Fear',\n 'Surprise',\n 'Surprise',\n 'Surprise',\n 'Neutral',\n 'Surprise',\n 'Fear',\n 'Surprise',\n 'Surprise',\n 'Surprise',\n 'Surprise',\n 'Surprise',\n 'Surprise',\n 'Surprise',\n 'Surprise',\n 'Neutral',\n 'Surprise',\n 'Surprise',\n 'Happy',\n 'Sad',\n 'Surprise',\n 'Surprise',\n 'Neutral',\n 'Surprise',\n 'Neutral',\n 'Surprise',\n 'Surprise',\n 'Surprise',\n 'Surprise',\n 'Fear',\n 'Surprise',\n 'Surprise',\n 'Neutral',\n 'Surprise',\n 'Fear',\n 'Surprise',\n 'Surprise',\n 'Neutral',\n 'Surprise',\n 'Surprise',\n 'Surprise',\n 'Surprise',\n 'Fear',\n 'Surprise',\n 'Fear',\n 'Neutral',\n 'Angry',\n 'Surprise',\n 'Surprise',\n 'Surprise',\n 'Fear',\n 'Sad',\n 'Surprise',\n 'Surprise',\n 'Surprise',\n 'Surprise',\n 'Surprise',\n 'Surprise',\n 'Surprise',\n 'Surprise',\n 'Surprise',\n 'Surprise',\n 'Surprise',\n 'Surprise',\n 'Surprise',\n 'Surprise',\n 'Fear',\n 'Surprise',\n 'Surprise',\n 'Fear',\n 'Surprise',\n 'Angry',\n 'Surprise',\n 'Fear',\n 'Surprise',\n 'Surprise',\n 'Fear',\n 'Surprise',\n 'Surprise',\n 'Surprise',\n 'Surprise',\n 'Surprise',\n 'Surprise',\n 'Fear',\n 'Surprise',\n 'Surprise',\n 'Surprise',\n 'Surprise',\n 'Surprise',\n 'Fear',\n 'Fear',\n 'Neutral',\n 'Surprise',\n 'Surprise',\n 'Surprise',\n 'Happy',\n 'Neutral',\n 'Neutral',\n 'Disgust',\n 'Neutral',\n 'Neutral',\n 'Sad',\n 'Neutral',\n 'Neutral',\n 'Neutral',\n 'Disgust',\n 'Neutral',\n 'Neutral',\n 'Neutral',\n 'Neutral',\n 'Neutral',\n 'Neutral',\n 'Neutral',\n 'Happy',\n 'Neutral',\n 'Neutral',\n 'Neutral',\n 'Angry',\n 'Neutral',\n 'Neutral',\n 'Neutral',\n 'Sad',\n 'Angry',\n 'Happy',\n 'Sad',\n 'Neutral',\n 'Neutral',\n 'Sad',\n 'Sad',\n 'Neutral',\n 'Surprise',\n 'Neutral',\n 'Neutral',\n 'Neutral',\n 'Happy',\n 'Neutral',\n 'Happy',\n 'Sad',\n 'Disgust',\n 'Neutral',\n 'Neutral',\n 'Neutral',\n 'Neutral',\n 'Neutral',\n 'Neutral',\n 'Neutral',\n 'Neutral',\n 'Neutral',\n 'Sad',\n 'Fear',\n 'Neutral',\n 'Neutral',\n 'Sad',\n 'Neutral',\n 'Disgust',\n 'Disgust',\n 'Neutral',\n 'Neutral',\n 'Neutral',\n 'Surprise',\n 'Neutral',\n 'Neutral',\n 'Happy',\n 'Disgust',\n 'Neutral',\n 'Sad',\n 'Neutral',\n 'Neutral',\n 'Neutral',\n 'Neutral',\n 'Happy',\n 'Disgust',\n 'Happy',\n 'Neutral',\n 'Neutral',\n 'Neutral',\n 'Neutral',\n 'Neutral',\n 'Happy',\n 'Neutral',\n 'Neutral',\n 'Neutral',\n 'Neutral',\n 'Neutral',\n 'Neutral',\n 'Neutral',\n 'Sad',\n 'Happy',\n 'Disgust',\n 'Neutral',\n 'Neutral',\n 'Happy',\n 'Angry',\n 'Neutral',\n 'Happy']", "_____no_output_____" ], [ "len(amazon_rekognition_labels)", "_____no_output_____" ], [ "import torch\nimport torchvision\nimport torchvision.transforms as transforms\nfrom torchvision.datasets.folder import pil_loader\nimport pickle\nPkl_Filename = \"P3ModelPyTorch.pkl\"\nwith open(Pkl_Filename, 'rb') as file: \n SavedModel = pickle.load(file)", "_____no_output_____" ], [ "transform_test = transforms.Compose([\n transforms.Resize((128,128)),\n transforms.ToTensor(),\n transforms.Normalize((0.4914, 0.4822, 0.4465), (0.2023, 0.1994, 0.2010)),\n ])", "_____no_output_____" ], [ "\ndef pytorch_model_prediction(im):\n newPic = transform_test(pil_loader(im))\n newPic = newPic.unsqueeze(0)\n outputNew = SavedModel(newPic)\n _, predicted = torch.max(outputNew.data, 1)\n if predicted == 0:\n return 'Anger'\n elif predicted == 1:\n return 'Disgust'\n elif predicted == 2:\n return 'Fear'\n elif predicted == 3:\n return 'Happy'\n elif predicted == 4:\n return 'Sad'\n elif predicted == 5:\n return 'Surprise'\n elif predicted == 6:\n return 'Neutral'\n \n# sm = torch.nn.Softmax(dim=1)\n# probabilities = sm(outputNew)\n# b = max(list(max(probabilities.data)))\n# print('{:f}'.format(b))", "_____no_output_____" ], [ "import os\n\nroot_path = '/Users/nvvankad/Documents/Personal/Masters/CSCE 5214 - Software Development for AI/P3 Project/test_data'\nfolders = ['0Angry', '1Disgust', '2Fear', '3Happy', '4Sad', '5Surprise', '6Neutral']\npytorch_predicted_labels = []\n\nfor folder in folders:\n for currentpath, folders, files in os.walk(os.path.join(root_path, folder)):\n for file in files:\n file_path = os.path.join(currentpath, file)\n if '.DS_Store' in file_path:\n continue\n pytorch_predicted_labels.append(pytorch_model_prediction(file_path))", "_____no_output_____" ], [ "len(pytorch_predicted_labels)", "_____no_output_____" ], [ "# Sample confusion matrix between true vs true.\nimport pandas as pd\nimport seaborn as sn\nimport matplotlib.pyplot as plt\n\ndata = {'amazon_rekogntion_predictions': amazon_rekognition_labels,\n 'pytorch_predictions': pytorch_predicted_labels\n }\n\ndf = pd.DataFrame(data, columns=['amazon_rekogntion_predictions','pytorch_predictions'])\nconfusion_matrix = pd.crosstab(df['amazon_rekogntion_predictions'], df['pytorch_predictions'], rownames=['Amazon Rekognition Predictions'], colnames=['Pytorch Model Predictions'])\n\nsn.heatmap(confusion_matrix, annot=True, fmt = \".0f\")\nplt.show()", "_____no_output_____" ] ] ]

[ "code" ]

[ [ "code", "code", "code", "code", "code", "code", "code", "code" ] ]

[ "Apache-2.0" ]

[ "Apache-2.0" ]

[ "Apache-2.0" ]

[ [ [ "<a href=\"https://colab.research.google.com/github/Nickamaes/Linear-Algebra_58109/blob/main/Practical_Lab_Exam_1_.ipynb\" target=\"_parent\"><img src=\"https://colab.research.google.com/assets/colab-badge.svg\" alt=\"Open In Colab\"/></a>", "_____no_output_____" ], [ "###Problem 1", "_____no_output_____" ] ], [ [ "import numpy as np\nA = np.array([[1,2,3],[4,5,6]])\nB = np.array([[1,2],[3,4],[5,6]]) \nC = np.array([[1,2,3],[4,5,6],[7,8,9]])\nD = np.array([[1,2],[3,4]]) \n\na = np.dot(A,B)\nb = np.add(D,D)\nc = 2*C\n\nprint(\"A. AB\")\nprint(a)\nprint(\"\\nB. D+D\")\nprint(b)\nprint(\"\\nC. 2*C\")\nprint(c)", "A. AB\n[[22 28]\n [49 64]]\n\nB. D+D\n[[2 4]\n [6 8]]\n\nC. 2*C\n[[ 2 4 6]\n [ 8 10 12]\n [14 16 18]]\n" ] ], [ [ "###Problem 2", "_____no_output_____" ] ], [ [ "import numpy as np\nX = np.array([5,3,-1])\n\nprint(X)\nprint('\\ntype of array', type(X))\nprint('size of array : ', X.size)\nprint('shape of array: ', X.shape)\nprint('dimension of array:', X.ndim)", "[ 5 3 -1]\n\ntype of array <class 'numpy.ndarray'>\nsize of array : 3\nshape of array: (3,)\ndimension of array: 1\n" ] ] ]

[ "markdown", "code", "markdown", "code" ]

[ [ "markdown", "markdown" ], [ "code" ], [ "markdown" ], [ "code" ] ]

[ "MIT" ]

[ "MIT" ]

[ "MIT" ]

[ [ [ "# Supervised Learning Decision Tree Classification Model Fitting", "_____no_output_____" ], [ "In this notebook, I will be applying the following techniques to predict Titanic survival:\n\n1. Decision Tree Classification (with default settings)\n2. Decision Tree Classification with the following optimal hyperparameters identified by using 5-fold Cross Validation\n * Max Depth\n * Min Samples Split\n * Min Samples Leaf\n * Max Features\n3. Random Forest\n\nLet's start by importing the necessary libraries.", "_____no_output_____" ] ], [ [ "import pandas as pd\nimport matplotlib.pyplot as plt\n%matplotlib inline\nimport numpy as np", "_____no_output_____" ] ], [ [ "Now import the train and test data sets. Take a quick glance at the data and get summary statistics for our variables.", "_____no_output_____" ] ], [ [ "train = pd.read_csv(\"data/new_train.csv\", index_col=0)\ntrain.head(5)", "_____no_output_____" ], [ "train.describe()", "_____no_output_____" ], [ "test = pd.read_csv(\"data/new_test.csv\", index_col=0)\ntest.head(5)", "_____no_output_____" ], [ "test.describe()", "_____no_output_____" ] ], [ [ "Feature columns from our data:", "_____no_output_____" ] ], [ [ "feature_cols = ['Pclass','LastName','Title','Sex','Age','SibSp','Parch','Fare','Embarked','TicketNumber','OtherName']", "_____no_output_____" ] ], [ [ "## Decision Tree Classification (with default settings)", "_____no_output_____" ], [ "Using the new_test and new_train CSV files, predict the Titanic survival with default model settings.", "_____no_output_____" ] ], [ [ "from sklearn import tree ", "_____no_output_____" ], [ "model = tree.DecisionTreeClassifier(random_state=42)\n# Prepare data for model fitting\nX = train.loc[:, feature_cols]\ny = train.Survived", "_____no_output_____" ], [ "# Fit a model\nmodel.fit(X, y)", "_____no_output_____" ], [ "X_test = test.loc[:, feature_cols]\nSurvived = model.predict(X_test)", "_____no_output_____" ], [ "test[\"Survived\"] = Survived", "_____no_output_____" ], [ "test.drop(['Pclass','LastName','Title','Sex','Age','SibSp','Parch','Fare','Embarked','TicketNumber','OtherName'],\n axis=1, inplace=True)", "_____no_output_____" ], [ "test.to_csv(\"data/submission.csv\", sep=',')", "_____no_output_____" ] ], [ [ "Result of the 1st submission is not bad:", "_____no_output_____" ], [ "", "_____no_output_____" ], [ "But I am curious if it can be improved if we applied better hyperparameters.", "_____no_output_____" ], [ "## Decision Tree Classification with optimal hyperparameters\n### Identified using 5-fold Cross Validation", "_____no_output_____" ], [ "Let's start by re-importing the train and test data sets so we have a clean, new data set.", "_____no_output_____" ] ], [ [ "train = pd.read_csv(\"data/new_train.csv\", index_col=0)\ntest = pd.read_csv(\"data/new_test.csv\", index_col=0)\nX = train.loc[:, feature_cols]\ny = train.Survived", "_____no_output_____" ] ], [ [ "### max_depth", "_____no_output_____" ], [ "The first parameter we will optimize is `max_depth`. Let's test 40 possible values from 1 to 40 for `max_depth` to identify the ideal value.", "_____no_output_____" ] ], [ [ "from sklearn.model_selection import train_test_split, cross_val_score\n\nmax_depth = np.arange(1,41,1)\nscores = []\n\n# split the data set into 80% training data and 20% test data\nXtrain, Xtest, ytrain, ytest = train_test_split(X,y,test_size=0.2)\n\nfor i in max_depth:\n model = tree.DecisionTreeClassifier(max_depth=i, random_state=42)\n avg_score = np.mean(cross_val_score(model, Xtrain, ytrain, cv=5))\n scores.append(avg_score)\n \n# plotting validation performance\nplt.plot(max_depth, scores)\nplt.xlabel(\"Max Depth\")\nplt.ylabel(\"Classification accuracy\")\n\n# Use best value of max_depth and report performance on the test split\nbest_max_depth = max_depth[np.argmax(scores)]\nbest_model = tree.DecisionTreeClassifier(max_depth=best_max_depth, random_state=42)\nbest_model.fit(Xtrain,ytrain)\n\n# reporting best accuracy\nprint(\"Value of max_depth with the best accuracy:\", best_max_depth)\nprint(\"Performance on test split:\", best_model.score(Xtest,ytest))", "Value of max_depth with the best accuracy: 3\nPerformance on test split: 0.8603351955307262\n" ] ], [ [ "### min_samples_split", "_____no_output_____" ], [ "The second parameter we will optimize is `min_samples_split`. Let's again test 40 possible values but this time, from 2 to 160 for `min_samples_split` to identify the ideal value.", "_____no_output_____" ] ], [ [ "min_samples_split = np.arange(2,160,4)\nscores = []\n\n# split the data set into 80% training data and 20% test data\nXtrain, Xtest, ytrain, ytest = train_test_split(X,y,test_size=0.2)\n\nfor i in min_samples_split:\n model = tree.DecisionTreeClassifier(min_samples_split=i, random_state=42)\n avg_score = np.mean(cross_val_score(model, Xtrain, ytrain, cv=5))\n scores.append(avg_score)\n \n# plotting validation performance\nplt.plot(min_samples_split, scores)\nplt.xlabel(\"Min Samples Split\")\nplt.ylabel(\"Classification accuracy\")\n\n# Use best value of min_samples_split and report performance on the test split\nbest_min_samples_split = min_samples_split[np.argmax(scores)]\nbest_model = tree.DecisionTreeClassifier(min_samples_split=best_min_samples_split, random_state=42)\nbest_model.fit(Xtrain,ytrain)\n\n# reporting best accuracy\nprint(\"Value of min_samples_split with the best accuracy:\", best_min_samples_split)\nprint(\"Performance on test split:\", best_model.score(Xtest,ytest))", "Value of min_samples_split with the best accuracy: 86\nPerformance on test split: 0.7988826815642458\n" ] ], [ [ "### min_samples_leaf", "_____no_output_____" ], [ "The third parameter we will optimize is `min_samples_leaf`. Similar to `max_depth`, let's test 40 possible values from 1 to 40 for `min_samples_leaf` to identify the ideal value.", "_____no_output_____" ] ], [ [ "min_samples_leaf = np.arange(1,41,1)\nscores = []\n\n# split the data set into 80% training data and 20% test data\nXtrain, Xtest, ytrain, ytest = train_test_split(X,y,test_size=0.2)\n\nfor i in min_samples_leaf:\n model = tree.DecisionTreeClassifier(min_samples_leaf=i, random_state=42)\n avg_score = np.mean(cross_val_score(model, Xtrain, ytrain, cv=5))\n scores.append(avg_score)\n \n# plotting validation performance\nplt.plot(min_samples_leaf, scores)\nplt.xlabel(\"Min Samples Leaf\")\nplt.ylabel(\"Classification accuracy\")\n\n# Use best value of min_samples_leaf and report performance on the test split\nbest_min_samples_leaf = min_samples_leaf[np.argmax(scores)]\nbest_model = tree.DecisionTreeClassifier(min_samples_leaf=best_min_samples_leaf, random_state=42)\nbest_model.fit(Xtrain,ytrain)\n\n# reporting best accuracy\nprint(\"Value of min_samples_leaf with the best accuracy:\", best_min_samples_leaf)\nprint(\"Performance on test split:\", best_model.score(Xtest,ytest))", "Value of min_samples_leaf with the best accuracy: 10\nPerformance on test split: 0.8324022346368715\n" ] ], [ [ "### max_features", "_____no_output_____" ], [ "The final parameter we will optimize is `max_features`", "_____no_output_____" ] ], [ [ "max_features = np.arange(1,len(feature_cols),1)\nscores = []\n\n# split the data set into 80% training data and 20% test data\nXtrain, Xtest, ytrain, ytest = train_test_split(X,y,test_size=0.2)\n\nfor i in max_features:\n model = tree.DecisionTreeClassifier(max_features=i, random_state=42)\n avg_score = np.mean(cross_val_score(model, Xtrain, ytrain, cv=5))\n scores.append(avg_score)\n \n# plotting validation performance\nplt.plot(max_features, scores)\nplt.xlabel(\"Max Features\")\nplt.ylabel(\"Classification accuracy\")\n\n# Use best value of max_features and report performance on the test split\nbest_max_features = max_features[np.argmax(scores)]\nbest_model = tree.DecisionTreeClassifier(max_features=best_max_features, random_state=42)\nbest_model.fit(Xtrain,ytrain)\n\n# reporting best accuracy\nprint(\"Value of max_features with the best accuracy:\", best_max_features)\nprint(\"Performance on test split:\", best_model.score(Xtest,ytest))", "Value of max_features with the best accuracy: 10\nPerformance on test split: 0.7821229050279329\n" ] ], [ [ "Let's take our _best_ optimized hyperparameters and create the _best_ model. Then, fit the model and make preditions.", "_____no_output_____" ] ], [ [ "best_model = tree.DecisionTreeClassifier(max_depth = best_max_depth,\n max_features = best_max_features,\n min_samples_split = best_min_samples_split,\n min_samples_leaf = best_min_samples_leaf,\n random_state=42)", "_____no_output_____" ], [ "# Fit a model\nbest_model.fit(X, y)", "_____no_output_____" ], [ "X_test = test.loc[:, feature_cols]\nSurvived = best_model.predict(X_test)", "_____no_output_____" ], [ "test[\"Survived\"] = Survived", "_____no_output_____" ] ], [ [ "Prepare the predictions for submission.", "_____no_output_____" ] ], [ [ "test.drop(['Pclass','LastName','Title','Sex','Age','SibSp','Parch','Fare','Embarked','TicketNumber','OtherName'],\n axis=1, inplace=True)", "_____no_output_____" ], [ "test.to_csv(\"data/submission.csv\", sep=',')", "_____no_output_____" ] ], [ [ "The results are an improvement:", "_____no_output_____" ], [ "", "_____no_output_____" ], [ "We can observe a 3.35% improvement of accuracy using optimized hyperparameters.", "_____no_output_____" ], [ "## Random Forest", "_____no_output_____" ], [ "A combination of multiple decision tree classifiers creates a very powerful classifier -- Random Forest. Let's predict using it now!", "_____no_output_____" ] ], [ [ "from sklearn.ensemble import RandomForestClassifier", "C:\\Users\\talha\\Anaconda3\\lib\\site-packages\\sklearn\\ensemble\\weight_boosting.py:29: DeprecationWarning: numpy.core.umath_tests is an internal NumPy module and should not be imported. It will be removed in a future NumPy release.\n from numpy.core.umath_tests import inner1d\n" ], [ "train = pd.read_csv(\"data/new_train.csv\", index_col=0)\ntest = pd.read_csv(\"data/new_test.csv\", index_col=0)\nX = train.loc[:, feature_cols]\ny = train.Survived", "_____no_output_____" ] ], [ [ "Let's do a multi-dimensional search for ideal hyperparameters.", "_____no_output_____" ] ], [ [ "# hyperparameters\nn_estimators = np.arange(2,100,4)\nmax_depth = np.arange(1,41,1)\nmax_features = np.arange(1,len(feature_cols),1)\n\n# split the data set into 80% training data and 20% test data\nXtrain, Xtest, ytrain, ytest = train_test_split(X,y,test_size=0.2)\n\n# accuracy\ntrain_accuracy = np.zeros((len(n_estimators),len(max_depth),len(max_features)))\ntest_accuracy = np.zeros((len(n_estimators),len(max_depth),len(max_features)))\n\n# search for best hyperparameters\nfor i,ne in enumerate(n_estimators):\n for j,md in enumerate(max_depth):\n for k,mf in enumerate(max_features):\n model = RandomForestClassifier()\n model.fit(Xtrain,ytrain)\n train_accuracy[i,j,k] = model.score(Xtrain,ytrain)\n test_accuracy[i,j,k] = model.score(Xtest,ytest)\n \n# Identify best value of hyperparameters\nideal_hp_index = np.unravel_index(np.argmax(test_accuracy, axis=None), test_accuracy.shape)\nn_est = n_estimators[ideal_hp_index[0]]\nmax_dep = max_depth[ideal_hp_index[1]]\nmax_feat = max_features[ideal_hp_index[2]]\n\n# reporting best accuracy\nprint(\"n_estimators:\",n_est)\nprint(\"max_depth:\",max_dep)\nprint(\"max_features:\",max_feat)\nprint(\"Generated a test accuracy of\", test_accuracy[ideal_hp_index])", "n_estimators: 46\nmax_depth: 37\nmax_features: 1\nGenerated a test accuracy of 0.8770949720670391\n" ], [ "best_random_forest = RandomForestClassifier(n_estimators=n_est,\n max_features = max_feat,\n max_depth = max_dep)", "_____no_output_____" ], [ "# Fit a model\nbest_random_forest.fit(X, y)", "_____no_output_____" ], [ "X_test = test.loc[:, feature_cols]\nSurvived = best_random_forest.predict(X_test)", "_____no_output_____" ], [ "test[\"Survived\"] = Survived", "_____no_output_____" ], [ "test.drop(['Pclass','LastName','Title','Sex','Age','SibSp','Parch','Fare','Embarked','TicketNumber','OtherName'],\n axis=1, inplace=True)", "_____no_output_____" ], [ "test.to_csv(\"data/submission.csv\", sep=',')", "_____no_output_____" ] ], [ [ "", "_____no_output_____" ] ] ]

[ "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown" ]

[ [ "markdown", "markdown" ], [ "code" ], [ "markdown" ], [ "code", "code", "code", "code" ], [ "markdown" ], [ "code" ], [ "markdown", "markdown" ], [ "code", "code", "code", "code", "code", "code", "code" ], [ "markdown", "markdown", "markdown", "markdown", "markdown" ], [ "code" ], [ "markdown", "markdown" ], [ "code" ], [ "markdown", "markdown" ], [ "code" ], [ "markdown", "markdown" ], [ "code" ], [ "markdown", "markdown" ], [ "code" ], [ "markdown" ], [ "code", "code", "code", "code" ], [ "markdown" ], [ "code", "code" ], [ "markdown", "markdown", "markdown", "markdown", "markdown" ], [ "code", "code" ], [ "markdown" ], [ "code", "code", "code", "code", "code", "code", "code" ], [ "markdown" ] ]

[ "MIT" ]

[ "MIT" ]

[ "MIT" ]

[ [ [ "#mount google drive to colab if not mounted before\nfrom google.colab import files\nfrom google.colab import drive \ndrive.mount('/content/drive')", "Mounted at /content/drive\n" ], [ "!wget https://miil-public-eu.oss-eu-central-1.aliyuncs.com/model-zoo/ASL/MS_COCO_TRresNet_M_224_81.8.pth", "--2021-05-13 14:15:46-- https://miil-public-eu.oss-eu-central-1.aliyuncs.com/model-zoo/ASL/MS_COCO_TRresNet_M_224_81.8.pth\nResolving miil-public-eu.oss-eu-central-1.aliyuncs.com (miil-public-eu.oss-eu-central-1.aliyuncs.com)... 47.254.187.26\nConnecting to miil-public-eu.oss-eu-central-1.aliyuncs.com (miil-public-eu.oss-eu-central-1.aliyuncs.com)|47.254.187.26|:443... connected.\nHTTP request sent, awaiting response... 200 OK\nLength: 118405727 (113M) [application/octet-stream]\nSaving to: ‘MS_COCO_TRresNet_M_224_81.8.pth’\n\nMS_COCO_TRresNet_M_ 100%[===================>] 112.92M 6.71MB/s in 15s \n\n2021-05-13 14:16:01 (7.70 MB/s) - ‘MS_COCO_TRresNet_M_224_81.8.pth’ saved [118405727/118405727]\n\n" ], [ "# must update before use!\n!pip install --upgrade --force-reinstall --no-deps kaggle\n\nfrom google.colab import files\n\nfiles.upload()\n\n! mkdir ~/.kaggle\n\n! cp kaggle.json ~/.kaggle/\n\n! chmod 600 ~/.kaggle/kaggle.json\n\n! kaggle datasets list\n\n! kaggle competitions download -c '2021s1comp5329assignment2'\n\n! unzip 2021s1comp5329assignment2.zip", "Collecting kaggle\n\u001b[?25l Downloading https://files.pythonhosted.org/packages/3a/e7/3bac01547d2ed3d308ac92a0878fbdb0ed0f3d41fb1906c319ccbba1bfbc/kaggle-1.5.12.tar.gz (58kB)\n\r\u001b[K |█████▋ | 10kB 21.0MB/s eta 0:00:01\r\u001b[K |███████████▏ | 20kB 27.3MB/s eta 0:00:01\r\u001b[K |████████████████▊ | 30kB 23.3MB/s eta 0:00:01\r\u001b[K |██████████████████████▎ | 40kB 17.8MB/s eta 0:00:01\r\u001b[K |███████████████████████████▉ | 51kB 8.7MB/s eta 0:00:01\r\u001b[K |████████████████████████████████| 61kB 2.9MB/s \n\u001b[?25hBuilding wheels for collected packages: kaggle\n Building wheel for kaggle (setup.py) ... \u001b[?25l\u001b[?25hdone\n Created wheel for kaggle: filename=kaggle-1.5.12-cp37-none-any.whl size=73053 sha256=845b2d5bf587ede2fa2f7551abf4bd0bf23549e4021098d6366244c6dce8ed3a\n Stored in directory: /root/.cache/pip/wheels/a1/6a/26/d30b7499ff85a4a4593377a87ecf55f7d08af42f0de9b60303\nSuccessfully built kaggle\nInstalling collected packages: kaggle\n Found existing installation: kaggle 1.5.12\n Uninstalling kaggle-1.5.12:\n Successfully uninstalled kaggle-1.5.12\nSuccessfully installed kaggle-1.5.12\n" ], [ "from google.colab import drive\ndrive.mount('/content/drive')", "Drive already mounted at /content/drive; to attempt to forcibly remount, call drive.mount(\"/content/drive\", force_remount=True).\n" ], [ "ls /content/COMP5329S1A2Dataset/data/17657.jpg", "/content/COMP5329S1A2Dataset/data/17657.jpg\n" ], [ "import sys\nsys.path.append('/content/drive/MyDrive/Colab Notebooks/ASL-main')", "_____no_output_____" ], [ "%pip install inplace-abn", "Collecting inplace-abn\n\u001b[?25l Downloading https://files.pythonhosted.org/packages/a4/27/c5791febcdd9af346b66dff19759898476f148177c02b02a72e07ca8aba0/inplace-abn-1.1.0.tar.gz (137kB)\n\r\u001b[K |██▍ | 10kB 22.4MB/s eta 0:00:01\r\u001b[K |████▊ | 20kB 25.8MB/s eta 0:00:01\r\u001b[K |███████▏ | 30kB 30.7MB/s eta 0:00:01\r\u001b[K |█████████▌ | 40kB 22.8MB/s eta 0:00:01\r\u001b[K |████████████ | 51kB 8.3MB/s eta 0:00:01\r\u001b[K |██████████████▎ | 61kB 8.3MB/s eta 0:00:01\r\u001b[K |████████████████▊ | 71kB 8.5MB/s eta 0:00:01\r\u001b[K |███████████████████ | 81kB 9.5MB/s eta 0:00:01\r\u001b[K |█████████████████████▌ | 92kB 10.3MB/s eta 0:00:01\r\u001b[K |███████████████████████▉ | 102kB 11.1MB/s eta 0:00:01\r\u001b[K |██████████████████████████▎ | 112kB 11.1MB/s eta 0:00:01\r\u001b[K |████████████████████████████▋ | 122kB 11.1MB/s eta 0:00:01\r\u001b[K |███████████████████████████████ | 133kB 11.1MB/s eta 0:00:01\r\u001b[K |████████████████████████████████| 143kB 11.1MB/s \n\u001b[?25hBuilding wheels for collected packages: inplace-abn\n Building wheel for inplace-abn (setup.py) ... \u001b[?25l\u001b[?25hdone\n Created wheel for inplace-abn: filename=inplace_abn-1.1.0-cp37-cp37m-linux_x86_64.whl size=3042384 sha256=358e9a13babe5b282aa1a9dc258478f4b91547d6ba9f2476cf51312cb33dc2c8\n Stored in directory: /root/.cache/pip/wheels/90/e6/ce/baadcff0441c600caa5874d4d3322a7909e724fb7abab21a15\nSuccessfully built inplace-abn\nInstalling collected packages: inplace-abn\nSuccessfully installed inplace-abn-1.1.0\n" ], [ "!python '/content/drive/MyDrive/Colab Notebooks/ASL-main/train1.py' -j 0 '/content/drive/MyDrive/Colab Notebooks/ASL-main/COMP5329S1A2Dataset'", "creating model...\ndone\n\nNumber of training images: 25500\nNumber of validation images: 4500\nlen(val_dataset)): 4500\nlen(train_dataset)): 25500\n24032.jpg\n/content/drive/MyDrive/Colab Notebooks/ASL-main/imagedata.py:85: UserWarning: To copy construct from a tensor, it is recommended to use sourceTensor.clone().detach() or sourceTensor.clone().detach().requires_grad_(True), rather than torch.tensor(sourceTensor).\n return torch.tensor(image, dtype=torch.float32), torch.tensor(targets, dtype=torch.float32)\n904.jpg\n15119.jpg\n16471.jpg\n6368.jpg\n21482.jpg\n21737.jpg\n23668.jpg\n10956.jpg\n12391.jpg\n10234.jpg\n15874.jpg\n20559.jpg\n8018.jpg\n3010.jpg\n23760.jpg\n3320.jpg\n21361.jpg\n18547.jpg\n17986.jpg\n23136.jpg\n9112.jpg\n17962.jpg\n2333.jpg\n9392.jpg\n20822.jpg\n11204.jpg\n12890.jpg\n12322.jpg\n2704.jpg\n17856.jpg\n16307.jpg\n13275.jpg\n2183.jpg\n15446.jpg\n12274.jpg\n10043.jpg\n8772.jpg\n21012.jpg\n20334.jpg\n24977.jpg\n23179.jpg\n18033.jpg\n15744.jpg\n12494.jpg\n7825.jpg\n24352.jpg\n16843.jpg\n649.jpg\n23423.jpg\n13477.jpg\n19057.jpg\n9601.jpg\n13948.jpg\n15107.jpg\n5656.jpg\n10329.jpg\n20903.jpg\n9549.jpg\n5859.jpg\n3155.jpg\n25143.jpg\n19760.jpg\n726.jpg\n25201.jpg\n2386.jpg\n4509.jpg\n23706.jpg\n15492.jpg\n13121.jpg\n774.jpg\n23791.jpg\n22256.jpg\n9126.jpg\n7843.jpg\n452.jpg\n11851.jpg\n13517.jpg\n7882.jpg\n1595.jpg\n5121.jpg\n5001.jpg\n21628.jpg\n6315.jpg\n24248.jpg\n2803.jpg\n15912.jpg\n13136.jpg\n1944.jpg\n15400.jpg\n6566.jpg\n6445.jpg\n6349.jpg\n8959.jpg\n25371.jpg\n15077.jpg\n10242.jpg\n14223.jpg\n15179.jpg\n22926.jpg\n1746.jpg\n20651.jpg\n11953.jpg\n11360.jpg\n12834.jpg\n13690.jpg\n12491.jpg\n15700.jpg\n5002.jpg\n19790.jpg\n2645.jpg\n17040.jpg\n7142.jpg\n13995.jpg\n13580.jpg\n24800.jpg\n24934.jpg\n20418.jpg\n9449.jpg\n538.jpg\n2888.jpg\n11082.jpg\n1249.jpg\n259.jpg\n10996.jpg\n11434.jpg\n1872.jpg\n7787.jpg\ntorch.Size([128, 3, 200, 200])\ntorch.Size([128, 20])\nTraceback (most recent call last):\n File \"/content/drive/MyDrive/Colab Notebooks/ASL-main/train1.py\", line 165, in <module>\n main()\n File \"/content/drive/MyDrive/Colab Notebooks/ASL-main/train1.py\", line 67, in main\n train_multi_label(model, train_loader, val_loader, args.lr)\n File \"/content/drive/MyDrive/Colab Notebooks/ASL-main/train1.py\", line 98, in train_multi_label\n loss = criterion(output, target)\n File \"/usr/local/lib/python3.7/dist-packages/torch/nn/modules/module.py\", line 889, in _call_impl\n result = self.forward(*input, **kwargs)\n File \"/content/drive/MyDrive/Colab Notebooks/ASL-main/src/loss_functions/losses.py\", line 33, in forward\n los_pos = y * torch.log(xs_pos.clamp(min=self.eps))\nRuntimeError: The size of tensor a (128) must match the size of tensor b (20) at non-singleton dimension 1\n" ], [ "import pands as pd\ndata_root_dir='/content/drive/MyDrive/Colab Notebooks/ASL-main/COMP5329S1A2Dataset'\n\ntrain_csv = pd.read_csv(os.path.join(data_root_dir, 'train.csv'))\n\n# read data form file, make binary array for all classes\ntrain_csv_image_names = train_csv[:]['ImageID']\n\ntrain_csv_label = train_csv[:]['Labels'].str.split(' ')\n\nprint(train_csv_label[:10])\n\nnp_train_csv_label = np.zeros((train_csv_label.shape[0],20), dtype=int)\nfor outter_it in range(train_csv_label.shape[0]):\n for inner_it in train_csv_label[outter_it]:\n np_train_csv_label[outter_it][int(inner_it)] = 1\n\nprint(np_train_csv_label[:10])", "_____no_output_____" ], [ "import torch\nmodel = torch.hub.load('pytorch/vision:v0.9.0', 'resnet18', pretrained=True)\nmodel = torch.hub.load('pytorch/vision:v0.9.0', 'resnet34', pretrained=True)\nmodel = torch.hub.load('pytorch/vision:v0.9.0', 'resnet50', pretrained=True)\nmodel = torch.hub.load('pytorch/vision:v0.9.0', 'googlenet', pretrained=True)", "Downloading: \"https://github.com/pytorch/vision/archive/v0.9.0.zip\" to /root/.cache/torch/hub/v0.9.0.zip\nDownloading: \"https://download.pytorch.org/models/resnet18-5c106cde.pth\" to /root/.cache/torch/hub/checkpoints/resnet18-5c106cde.pth\n" ], [ "# https://github.com/jiangqy/Customized-DataLoader-pytorch/blob/master/multi_label_classifier.py\n# https://debuggercafe.com/multi-label-image-classification-with-pytorch-and-deep-learning/", "_____no_output_____" ] ] ]

[ "code" ]

[ [ "code", "code", "code", "code", "code", "code", "code", "code", "code", "code", "code" ] ]

[ "MIT" ]

[ "MIT" ]

[ "MIT" ]

[ [ [ "<a href=\"https://colab.research.google.com/github/jacKlinc/german_char_recogniser/blob/main/mdl_german_char_classifier.ipynb\" target=\"_parent\"><img src=\"https://colab.research.google.com/assets/colab-badge.svg\" alt=\"Open In Colab\"/></a>", "_____no_output_____" ] ], [ [ "#hide\n!pip install -Uqq fastbook\nimport fastbook\nfastbook.setup_book()", "_____no_output_____" ], [ "from fastbook import *\nfrom fastai.vision.all import *", "_____no_output_____" ], [ "path = Path('gdrive/MyDrive/Colab Notebooks')\nPath.BASE_PATH = path", "_____no_output_____" ], [ "a = (path/'A').ls().sorted()\nb = (path/'B').ls().sorted()", "_____no_output_____" ], [ "len(b)", "_____no_output_____" ], [ "img = Image.open(a[100])\nimg", "_____no_output_____" ], [ "img_t = tensor(img)", "_____no_output_____" ], [ "df = pd.DataFrame(img_t)\ndf.style.set_properties(**{'font-size': '6pt'}).background_gradient('Greys')", "_____no_output_____" ] ], [ [ "The cell above shows how the image is simply made of black and white and pixels; the darker the pixel, the higher the number in the cell.", "_____no_output_____" ], [ "## Pixel Similarity\nFind the mean shape of A and comapre each new image to that.", "_____no_output_____" ] ], [ [ "# get first 64 for testing\na_tensor = [tensor(Image.open(x)) for x in a[:64]]\nb_tensor = [tensor(Image.open(x)) for x in b[:64]]", "_____no_output_____" ] ], [ [ "To make this into one big matrix a neural network can use we need to stack all the tensors on top of each other.", "_____no_output_____" ] ], [ [ "stacked_a = torch.stack(a_tensor)\nstacked_b = torch.stack(b_tensor)", "_____no_output_____" ] ], [ [ "The data must be converted from a 0-255 scale to 0-1 for the neural network. The tensors are integers so they must be converted to floats too.", "_____no_output_____" ] ], [ [ "stacked_a = stacked_a.float()/255\nstacked_b = stacked_b.float()/255", "_____no_output_____" ] ], [ [ "Find the mean of each matrix.", "_____no_output_____" ] ], [ [ "mean_a = stacked_a.mean(0)\nmean_b = stacked_b.mean(0)", "_____no_output_____" ] ], [ [ "Getting the distance from the ideal (mean) character is going to determine how it is classified. There are two methods we can use:\n\n1. L1 norm - absolute mean value. This subtracts the mean from each value in the tensor and makes it postive.\n1. L2 norm - Root Mean Squared Error (RMSE). This subtracts the mean in the tensor but squares the difference. This makes for smaller errors having little weight and larger ones having quite an effect.\n\n", "_____no_output_____" ] ], [ [ "a0 = stacked_a[0]\nshow_image(a0)", "_____no_output_____" ], [ "dist_a_abs = (a0 - mean_a).abs().mean()\ndist_a_rmse = ((a0 - mean_a)**2).mean().sqrt()\ndist_a_abs, dist_a_rmse", "_____no_output_____" ], [ "dist_b_abs = (a0 - mean_b).abs().mean()\ndist_b_rmse = ((a0 - mean_b)**2).mean().sqrt()\ndist_b_abs, dist_b_rmse", "_____no_output_____" ] ], [ [ "Using either of the above approaches, the average distance between the \"a\" example and the ideal \"a\" is less than that between the \"b\" example. Either can be used to classify. Defining a function for each might be the most convenient way to proceed.", "_____no_output_____" ] ], [ [ "def abs_dist(ex, mean):\n return (ex - mean).abs().mean((-1, -2))\n\ndef rmse_dist(ex, mean):\n return ((ex - mean)**2).mean().sqrt()\n\nabs_dist(a0, mean_b), rmse_dist(a0, mean_b)", "_____no_output_____" ] ], [ [ "Now the error between an example and the mean can be found easily. Defining a way to compare the error to other character is the next step.\n\nThe code below checks if the absolute mean error between an \"a\" example and the mean of \"a\" is less than that of a \"b\" example.", "_____no_output_____" ] ], [ [ "abs_dist(a0, mean_a) < abs_dist(a0, mean_b)", "_____no_output_____" ], [ "def is_a(ex_a):\n return (abs_dist(ex_a, mean_a) < abs_dist(ex_a, mean_b)).float()\n\nis_a(a0) ", "_____no_output_____" ] ], [ [ "### Create Validation Set\nGather validation data to measure accuracy of classification.", "_____no_output_____" ] ], [ [ "valid_stacked_a = torch.stack([tensor(Image.open(x)) \n for x in a[65:128]]).float()/255\n \nvalid_stacked_b = torch.stack([tensor(Image.open(x)) \n for x in b[65:128]]).float()/255\n\nvalid_stacked_a.shape, valid_stacked_b.shape", "_____no_output_____" ], [ "accuracy_a = is_a(valid_stacked_a) .mean()\naccuracy_b = (1 - is_a(valid_stacked_b)).mean()\n\naccuracy_a, accuracy_b", "_____no_output_____" ] ], [ [ "Next we need to find the distance from each value in the validation matrix from the ideal \"a\". This could be acheived by looping over the each value but this would be inefficient. \n\nLuckily, broadcasting exists, this is where PyTorch has tensors of different ranks and it expands the smaller one (```mean_a```) to the size of the larger one (```valid_stacked_a```) so that each value is operated on.\n\nThis makes things much faster because the operation is passed directly through C or if running on a GPU, it can be millions of times faster.", "_____no_output_____" ] ], [ [ "valid_a_dist = abs_dist(valid_stacked_a, mean_a)\nvalid_a_dist, valid_a_dist.shape", "_____no_output_____" ] ], [ [ "### Stochastic Gradient Descent (SGD)\nAlgorithm used to optimise weights and parameters so that the loss function is minimised. Gradient descent wants to find the lowest point on the graph so that the losses are also low.\n\n**7 Steps:**\n1. Init weights\n1. Predict character\n1. Calculate loss\n1. Calculate gradient\n1. Make step\n1. Repeat 2.\n1. End when converged or out of time\n\nTo calculate this gradient, we need to find the derivative of the loss function. The derivative will basically find the slope of the curve for a given weight, each value for the gradient will return in a matrix.", "_____no_output_____" ] ], [ [ "xt = tensor(3.).requires_grad_()", "_____no_output_____" ] ], [ [ "The ```requires_grad_()``` function marks the variable so that when it is used in future it knows that the gradient needs to be calculated.", "_____no_output_____" ] ], [ [ "def f(x): return x**2\n\nyt = f(xt)\nyt", "_____no_output_____" ] ], [ [ "Calculate the gradients now", "_____no_output_____" ] ], [ [ "yt.backward()", "_____no_output_____" ] ], [ [ "The ```.backward()``` function applies the backpropagation algorithm to each layer of the network", "_____no_output_____" ] ], [ [ "xt.grad", "_____no_output_____" ] ], [ [ "The gradients only tell us the slope of our weights and don't really provide anything actionable to use. This is where steps are used.", "_____no_output_____" ], [ "### Steps\nThis step of the process uses the gradients to attempt to lower the loss function by \"jumping\" down the curve. The size of the jump is called the learning late ```lr```, the bigger it is the more risky it is, as it could overshoot and increase the loss; too small and training the algorithm would take too long and would be expensive in real-world applications.\n\nThe formula can be described here:\n\n``` w = w - gradient(w) * lr```\n\nwhere w is each weight.\n\nSteps are taken until convergence or until you run out of time. ", "_____no_output_____" ], [ "### Sigmoid Function\nThe issue with above approach is the output will be linear meaning I get a number back when I predict, this isn't much use for telling me whether it's an \"a\" or a \"b\". Enter sigmoid, this gets a function and polarises the output so that a yes/no, a/b result is found.", "_____no_output_____" ] ], [ [ "def sigmoid(x):\n return 1 / (1 + torch.exp(-x))\n\nplot_function(torch.sigmoid, title='Sigmoid', min=-4, max=4)", "_____no_output_____" ] ], [ [ "After the midpoint (0.5) all values below are 0 and all above are 1. This can now be used in the loss function.", "_____no_output_____" ] ], [ [ "def char_loss(preds, targets):\n preds = preds.sigmoid()\n # return mean of either: preds or (1-preds)\n return torch.where(preds == 1, 1 - preds, preds).mean()", "_____no_output_____" ] ], [ [ "### Mini-Batches\nThe loss function has been defined, we're ready to run gradient descent on the data. Well maybe not there's a few more things to iron out. The issue with running gradient descent on all our data is that would take quite a while; instead we could run it on each item, this would be much faster. The issue with operated on each example is that it doesn't account for other examples.\n\n\nThe Goldielox solution is to use mini-batches, it's faster than calculating the loss for the entire set but more accurate than doing it on each datum. FastAI provides the ```DataLoader``` to handle this.", "_____no_output_____" ] ], [ [ "col = range(30)\n# bs=batch size\ndl = DataLoader(col, bs=5, shuffle=True)\nlist(dl)", "_____no_output_____" ] ], [ [ "## Implementation\n", "_____no_output_____" ] ], [ [ "path = Path('gdrive/MyDrive/Colab Notebooks/characters')\nPath.BASE_PATH = path", "_____no_output_____" ], [ "fns = get_image_files(path)\nfns", "_____no_output_____" ], [ "failed = verify_images(fns)\nfailed", "_____no_output_____" ] ], [ [ "Delete failed images", "_____no_output_____" ] ], [ [ "failed.map(Path.unlink);", "_____no_output_____" ] ], [ [ "### Data Curation", "_____no_output_____" ] ], [ [ "characters = DataBlock(\n blocks=(ImageBlock, CategoryBlock), \n get_items=get_image_files, \n splitter=RandomSplitter(valid_pct=0.2, seed=42),\n get_y=parent_label)", "_____no_output_____" ] ], [ [ "This defines how the data will be retreived and organised. The ```splitter``` divides the data into the training and validation sets.", "_____no_output_____" ] ], [ [ "dls = characters.dataloaders(path)", "_____no_output_____" ], [ "dls.valid.show_batch(max_n=4, nrows=1)", "_____no_output_____" ] ], [ [ "### Model Training", "_____no_output_____" ] ], [ [ "characters = characters.new(\n # item_tfms=RandomResizedCrop(224, min_scale=0.5),\n batch_tfms=aug_transforms())\ndls = characters.dataloaders(path)", "_____no_output_____" ], [ "learn = cnn_learner(dls, resnet18, metrics=error_rate)\nlearn.fine_tune(5)", "Downloading: \"https://download.pytorch.org/models/resnet18-5c106cde.pth\" to /root/.cache/torch/hub/checkpoints/resnet18-5c106cde.pth\n" ], [ "interp = ClassificationInterpretation.from_learner(learn)\ninterp.plot_confusion_matrix()", "_____no_output_____" ], [ "interp.plot_top_losses(3, nrows=1)", "_____no_output_____" ] ], [ [ "### Model Inference\nExport model to app.", "_____no_output_____" ] ], [ [ "learn.export()", "_____no_output_____" ], [ "path = Path()\npath.ls(file_exts='.pkl')", "_____no_output_____" ], [ "%mv export.pkl gdrive/MyDrive/Colab\\ Notebooks/characters", "_____no_output_____" ], [ "%ls gdrive/MyDrive/Colab\\ Notebooks/characters", "\u001b[0m\u001b[01;34mA\u001b[0m/ \u001b[01;34mB\u001b[0m/ export.pkl\n" ] ] ]

[ "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code" ]

[ [ "markdown" ], [ "code", "code", "code", "code", "code", "code", "code", "code" ], [ "markdown", "markdown" ], [ "code" ], [ "markdown" ], [ "code" ], [ "markdown" ], [ "code" ], [ "markdown" ], [ "code" ], [ "markdown" ], [ "code", "code", "code" ], [ "markdown" ], [ "code" ], [ "markdown" ], [ "code", "code" ], [ "markdown" ], [ "code", "code" ], [ "markdown" ], [ "code" ], [ "markdown" ], [ "code" ], [ "markdown" ], [ "code" ], [ "markdown" ], [ "code" ], [ "markdown" ], [ "code" ], [ "markdown", "markdown", "markdown" ], [ "code" ], [ "markdown" ], [ "code" ], [ "markdown" ], [ "code" ], [ "markdown" ], [ "code", "code", "code" ], [ "markdown" ], [ "code" ], [ "markdown" ], [ "code" ], [ "markdown" ], [ "code", "code" ], [ "markdown" ], [ "code", "code", "code", "code" ], [ "markdown" ], [ "code", "code", "code", "code" ] ]

[ "MIT" ]

[ "MIT" ]

[ "MIT" ]

[ [ [ "import numpy as np\nimport pandas as pd\nimport multiprocessing as mp\nfrom multiprocessing import Pool, freeze_support\nfrom itertools import repeat\nimport time\n\ndef softT(m, t, zeta):\n \"\"\"\n Soft-Threshold for solving univariate lasso.\n \"\"\"\n # m: float; m_j in the thesis\n # t: float; truncation threshold\n # zeta: float; penalty factor, the same as the `lambda` in the thesis.\n if (zeta < abs(m)):\n z = (-m - zeta) / (2 * t) if (m < 0) else (-m + zeta) / (2 * t)\n else:\n z = 0\n return z\n\ndef unpack_args(X, Y, beta, beta0, zeta, chunks = 4):\n args_iterator = []\n n, _ = X.shape\n chunk_size = np.int(n/chunks) + 1\n for i in range(chunks):\n args_iterator.append((X[i*chunk_size:(i+1)*chunk_size], Y[i*chunk_size:(i+1)*chunk_size], beta, beta0, zeta))\n return args_iterator\n\ndef calculate_object_function(X, Y, beta, beta0, zeta):\n \"\"\"\n Calculate objetc function value.\n @Input:\n X: dataframe; n * p\n Y: dataframe; n * 1\n beta: 2-d np.array; p*1\n beta0: float\n zeta: float\n @Output:\n fval: 2-d np.array; n*1\n \"\"\"\n X_by_beta_plus_beta0 = np.matmul(X, beta) + beta0\n log_likelihood = np.sum(Y * X_by_beta_plus_beta0 - np.log(1 + np.exp(X_by_beta_plus_beta0)))\n fval = (-log_likelihood) / (2 * n) + zeta * np.sum(np.abs(beta)) \n return fval\n\n\nif __name__ == \"__main__\":\n data = pd.read_csv(\"./toyexample.csv\")\n data = data.drop([\"Unnamed: 0\"], axis=1)\n X = data.drop('y', axis=1)\n Y = data.iloc[:,-1].values.reshape(-1,1)\n n, p = X.shape\n # rand_seed = np.random.RandomState(1234)\n # n, p = 1000, 100\n # X = rand_seed.beta(1,2,size=(n, p))\n # Y = rand_seed.binomial(n=1, p=0.2, size=n).reshape(-1,1)\n beta = np.ones(p).reshape(-1,1)\n beta0 = 1\n zeta = 0.1\n parallel = False\n print(\"Shape of X is n={}, p={}\".format(n,p))\n serial_start = time.time()\n serila_result = calculate_object_function(X, Y, beta, beta0, zeta)\n serial_end = time.time()\n print(\"Serial took: [{}] s\".format(serial_end - serial_start))\n if parallel:\n num_cores = mp.cpu_count()\n pool_start = time.time()\n args = unpack_args(X, Y, beta, beta0, zeta, chunks=10)\n t = time.time()\n with Pool(processes = num_cores) as pool:\n pool_result = pool.starmap(calculate_object_function, args)\n tt = time.time()\n pool_end = time.time()\n print(\"Pool took: [{}] s\".format(pool_end - pool_start))\n \n try:\n print(\"Difference betwwen two results: {}\".format((np.sum(pool_result) - np.sum(serila_result))/n))\n print(tt-t)\n except NameError:\n pass", "_____no_output_____" ] ] ]

[ "code" ]

[ [ "code" ] ]

[ "BSD-3-Clause-Clear" ]

[ "BSD-3-Clause-Clear" ]

[ "BSD-3-Clause-Clear" ]

[ [ [ "# Managing assignment files", "_____no_output_____" ], [ "Distributing assignments to students and collecting them can be a logistical nightmare. If you are running nbgrader on a server, some of this pain can be relieved by relying on nbgrader's built-in functionality for releasing and collecting assignments on the instructor's side, and fetching and submitting assignments on the student's side.", "_____no_output_____" ] ], [ [ ".. contents:: Table of Contents\n :depth: 2", "_____no_output_____" ] ], [ [ "## Setting up the exchange", "_____no_output_____" ], [ "After an assignment has been created using `nbgrader assign`, the instructor must actually release that assignment to students. If the class is being taught on a single filesystem, then the instructor may use `nbgrader release` to copy the assignment files to a shared location on the filesystem for students to then download.\n\nFirst, we must specify a few configuration options. To do this, we'll create a `nbgrader_config.py` file that will get automatically loaded when we run `nbgrader`:", "_____no_output_____" ] ], [ [ "%%file nbgrader_config.py\n\nc = get_config()\n\nc.Exchange.course_id = \"example_course\"\nc.Exchange.root = \"/tmp/exchange\"", "Writing nbgrader_config.py\n" ] ], [ [ "In the config file, we've specified the \"exchange\" directory to be `/tmp/exchange`. This directory must exist before running `nbgrader`, and it *must* be readable and writable by all users, so we'll first create it and configure the appropriate permissions:", "_____no_output_____" ] ], [ [ "%%bash\n\n# remove existing directory, so we can start fresh for demo purposes\nrm -rf /tmp/exchange\n\n# create the exchange directory, with write permissions for everyone\nmkdir /tmp/exchange\nchmod ugo+rw /tmp/exchange", "_____no_output_____" ] ], [ [ "## Releasing assignments", "_____no_output_____" ] ], [ [ ".. seealso::\n\n :doc:`creating_and_grading_assignments`\n Details on generating assignments\n\n :doc:`/command_line_tools/nbgrader-release`\n Command line options for ``nbgrader release``\n\n :doc:`/command_line_tools/nbgrader-list`\n Command line options for ``nbgrader list``\n\n :doc:`philosophy`\n More details on how the nbgrader hierarchy is structured.\n\n :doc:`/configuration/config_options`\n Details on ``nbgrader_config.py``", "_____no_output_____" ] ], [ [ "### From the formgrader", "_____no_output_____" ], [ "Using the formgrader extension, you may release assignments by clicking on the \"release\" button:\n\n\n\n**Note** that for the \"release\" button to become available, the `course_id` option must be set in `nbgrader_config.py`.\nOnce completed, you will see a pop-up window with log output:\n\n\n\nIf you decide you want to \"un-release\" an assignment, you may do so by clicking again on the \"release\" button (which is now an \"x\"). **However, note that students who have already downloaded the assignment will still have access to their downloaded copy. Unreleasing an assignment only prevents more students from downloading it.**\n\n", "_____no_output_____" ], [ "### From the command line", "_____no_output_____" ], [ "Now that we have the directory created, we can actually run `nbgrader release` (and as with the other nbgrader commands for instructors, this must be run from the root of the course directory):", "_____no_output_____" ] ], [ [ "%%bash\n\nnbgrader release \"ps1\"", "[ReleaseApp | INFO] Source: /Users/jhamrick/project/tools/nbgrader/nbgrader/docs/source/user_guide/release/./ps1\n[ReleaseApp | INFO] Destination: /tmp/exchange/example_course/outbound/ps1\n[ReleaseApp | INFO] Released as: example_course ps1\n" ] ], [ [ "Finally, you can verify that the assignment has been appropriately released by running the `nbgrader list` command:", "_____no_output_____" ] ], [ [ "%%bash\n\nnbgrader list", "[ListApp | INFO] Released assignments:\n[ListApp | INFO] example_course ps1\n" ] ], [ [ "Note that there should only ever be *one* instructor who runs the `nbgrader release` and `nbgrader collect` commands (and there should probably only be one instructor -- the same instructor -- who runs `nbgrader assign`, `nbgrader autograde` and the formgrader as well). However this does not mean that only one instructor can do the grading, it just means that only one instructor manages the assignment files. Other instructors can still perform grading by accessing the notebook where the formgrader is running.", "_____no_output_____" ] ], [ [ ".. _fetching-assignments:", "_____no_output_____" ] ], [ [ "## Fetching assignments", "_____no_output_____" ] ], [ [ ".. seealso::\n\n :doc:`/command_line_tools/nbgrader-fetch`\n Command line options for ``nbgrader fetch``\n\n :doc:`/command_line_tools/nbgrader-list`\n Command line options for ``nbgrader list``\n\n :doc:`/configuration/config_options`\n Details on ``nbgrader_config.py``", "_____no_output_____" ] ], [ [ "From the student's perspective, they can list what assignments have been released, and then fetch a copy of the assignment to work on. First, we'll create a temporary directory to represent the student's home directory:", "_____no_output_____" ] ], [ [ "%%bash\n\n# remove the fake student home directory if it exists, for demo purposes\nrm -rf /tmp/student_home\n\n# create the fake student home directory and switch to it\nmkdir /tmp/student_home", "_____no_output_____" ] ], [ [ "If you are not using the default exchange directory (as is the case here), you will additionally need to provide your students with a configuration file that sets the appropriate directory for them:", "_____no_output_____" ] ], [ [ "%%file /tmp/student_home/nbgrader_config.py\n\nc = get_config()\nc.Exchange.root = '/tmp/exchange'\nc.Exchange.course_id = \"example_course\"", "Writing /tmp/student_home/nbgrader_config.py\n" ] ], [ [ "### From the notebook dashboard", "_____no_output_____" ] ], [ [ ".. warning::\n\n The \"Assignment List\" extension is not fully compatible with multiple\n courses on the same server. Please see :ref:`multiple-classes` for details.\n\nAlternatively, students can fetch assignments using the assignment list notebook server extension. You must have installed the extension by following the instructions :doc:`here </user_guide/installation>`, after which you should see an \"Assignments\" tab in dashboard:", "_____no_output_____" ] ], [ [ "", "_____no_output_____" ], [ "The image above shows that there has been one assignment released (\"ps1\") for the class \"example_course\". To get this assignment, students can click the \"Fetch\" button (analogous to running `nbgrader fetch ps1 --course example_course`. **Note: this assumes nbgrader is always run from the root of the notebook server, which on JupyterHub is most likely the root of the user's home directory.**\n\nAfter the assignment is fetched, it will appear in the list of \"Downloaded assignments\":\n\n", "_____no_output_____" ], [ "Students can click on the name of the assignment to expand it and see all the notebooks in the assignment:\n\n\n\nClicking on a particular notebook will open it in a new tab in the browser.", "_____no_output_____" ], [ "### From the command line", "_____no_output_____" ], [ "From the student's perspective, they can see what assignments have been released using `nbgrader list`, and passing the name of the class:", "_____no_output_____" ] ], [ [ "%%bash\nexport HOME=/tmp/student_home && cd $HOME\n\nnbgrader list", "[ListApp | INFO] Released assignments:\n[ListApp | INFO] example_course ps1\n" ] ], [ [ "They can then fetch an assignment for that class using `nbgrader fetch` and passing the name of the class and the name of the assignment:", "_____no_output_____" ] ], [ [ "%%bash\nexport HOME=/tmp/student_home && cd $HOME\n\nnbgrader fetch \"ps1\"", "[FetchApp | INFO] Source: /tmp/exchange/example_course/outbound/ps1\n[FetchApp | INFO] Destination: /private/tmp/student_home/ps1\n[FetchApp | INFO] Fetched as: example_course ps1\n" ] ], [ [ "Note that running `nbgrader fetch` copies the assignment files from the exchange directory to the local directory, and therefore can be used from any directory:", "_____no_output_____" ] ], [ [ "%%bash\n\nls -l \"/tmp/student_home/ps1\"", "total 40\n-rw-r--r-- 1 jhamrick wheel 5733 Apr 22 15:29 jupyter.png\n-rw-r--r-- 1 jhamrick wheel 8126 Apr 22 15:29 problem1.ipynb\n-rw-r--r-- 1 jhamrick wheel 2318 Apr 22 15:29 problem2.ipynb\n" ] ], [ [ "Additionally, the `nbgrader fetch` (as well as `nbgrader submit`) command also does not rely on having access to the nbgrader database -- the database is only used by instructors.", "_____no_output_____" ], [ "## Submitting assignments", "_____no_output_____" ] ], [ [ ".. seealso::\n\n :doc:`/command_line_tools/nbgrader-submit`\n Command line options for ``nbgrader fetch``\n\n :doc:`/command_line_tools/nbgrader-list`\n Command line options for ``nbgrader list``\n\n :doc:`/configuration/config_options`\n Details on ``nbgrader_config.py``", "_____no_output_____" ] ], [ [ "### From the notebook dashboard", "_____no_output_____" ] ], [ [ ".. warning::\n\n The \"Assignment List\" extension is not fully compatible with multiple\n courses on the same server. Please see :ref:`multiple-classes` for details.\n\nAlternatively, students can submit assignments using the assignment list notebook server extension. You must have installed the extension by following the instructions `here <https://github.com/jupyter/nbgrader>`__. Students must have also downloaded the assignments (see :ref:`fetching-assignments`).", "_____no_output_____" ] ], [ [ "After students have worked on the assignment for a while, but before submitting, they can validate that their notebooks pass the tests by clicking the \"Validate\" button (analogous to running `nbgrader validate`). If any tests fail, they will see a warning:\n\n", "_____no_output_____" ], [ "If there are no errors, they will see that the validation passes:\n\n", "_____no_output_____" ] ], [ [ ".. note::\n\n If the notebook has been released with hidden tests removed from the source version\n (see :ref:`autograder-tests-cell-hidden-tests`) then this validation is only done against the tests the students can\n see in the release version.", "_____no_output_____" ] ], [ [ "Once students have validated all the notebooks, they can click the \"Submit\" button to submit the assignment (analogous to running `nbgrader submit ps1 --course example_course`). Afterwards, it will show up in the list of submitted assignments (and also still in the list of downloaded assignments):\n\n", "_____no_output_____" ], [ "Students may submit an assignment as many times as they'd like. All copies of a submission will show up in the submitted assignments list, and when the instructor collects the assignments, they will get the most recent version of the assignment:\n\n", "_____no_output_____" ], [ "Similarly, if the ``strict`` option (in the student's ``nbgrader_config.py`` file) is set to ``True``, the students will not be able to submit an assignment with missing notebooks (for a given assignment):\n\n", "_____no_output_____" ], [ "### From the command line", "_____no_output_____" ], [ "First, as a reminder, here is what the student's `nbgrader_config.py` file looks like:", "_____no_output_____" ] ], [ [ "%%bash\n\ncat /tmp/student_home/nbgrader_config.py", "\nc = get_config()\nc.Exchange.root = '/tmp/exchange'\nc.Exchange.course_id = \"example_course\"" ] ], [ [ "After working on an assignment, the student can submit their version for grading using `nbgrader submit` and passing the name of the assignment and the name of the class:", "_____no_output_____" ] ], [ [ "%%bash\nexport HOME=/tmp/student_home && cd $HOME\n\nnbgrader submit \"ps1\"", "[SubmitApp | INFO] Source: /private/tmp/student_home/ps1\n[SubmitApp | INFO] Destination: /tmp/exchange/example_course/inbound/jhamrick+ps1+2018-04-22 14:29:40.397476 UTC\n[SubmitApp | INFO] Submitted as: example_course ps1 2018-04-22 14:29:40.397476 UTC\n" ] ], [ [ "Note that \"the name of the assignment\" really corresponds to \"the name of a folder\". It just happens that, in our current directory, there is a folder called \"ps1\":", "_____no_output_____" ] ], [ [ "%%bash\nexport HOME=/tmp/student_home && cd $HOME\n\nls -l \"/tmp/student_home\"", "total 8\ndrwxr-xr-x 3 jhamrick wheel 96 Apr 22 15:29 Library\n-rw-r--r-- 1 jhamrick wheel 91 Apr 22 15:29 nbgrader_config.py\ndrwxr-xr-x 5 jhamrick wheel 160 Apr 22 15:29 ps1\n" ] ], [ [ "Students can see what assignments they have submitted using `nbgrader list --inbound`:", "_____no_output_____" ] ], [ [ "%%bash\nexport HOME=/tmp/student_home && cd $HOME\n\nnbgrader list --inbound", "[ListApp | INFO] Submitted assignments:\n[ListApp | INFO] example_course jhamrick ps1 2018-04-22 14:29:40.397476 UTC\n" ] ], [ [ "Importantly, students can run `nbgrader submit` as many times as they want, and all submitted copies of the assignment will be preserved:", "_____no_output_____" ] ], [ [ "%%bash\nexport HOME=/tmp/student_home && cd $HOME\n\nnbgrader submit \"ps1\"", "[SubmitApp | INFO] Source: /private/tmp/student_home/ps1\n[SubmitApp | INFO] Destination: /tmp/exchange/example_course/inbound/jhamrick+ps1+2018-04-22 14:29:43.070290 UTC\n[SubmitApp | INFO] Submitted as: example_course ps1 2018-04-22 14:29:43.070290 UTC\n" ] ], [ [ "We can see all versions that have been submitted by again running `nbgrader list --inbound`:", "_____no_output_____" ] ], [ [ "%%bash\nexport HOME=/tmp/student_home && cd $HOME\n\nnbgrader list --inbound", "[ListApp | INFO] Submitted assignments:\n[ListApp | INFO] example_course jhamrick ps1 2018-04-22 14:29:40.397476 UTC\n[ListApp | INFO] example_course jhamrick ps1 2018-04-22 14:29:43.070290 UTC\n" ] ], [ [ "Note that the `nbgrader submit` (as well as `nbgrader fetch`) command also does not rely on having access to the nbgrader database -- the database is only used by instructors.", "_____no_output_____" ], [ "``nbgrader`` requires that the submitted notebook names match the released notebook names for each assignment. For example if a student were to rename one of the given assignment notebooks:", "_____no_output_____" ] ], [ [ "%%bash\nexport HOME=/tmp/student_home && cd $HOME\n\n# assume the student renamed the assignment file\nmv ps1/problem1.ipynb ps1/myproblem1.ipynb\n\nnbgrader submit \"ps1\"", "[SubmitApp | INFO] Source: /private/tmp/student_home/ps1\n[SubmitApp | INFO] Destination: /tmp/exchange/example_course/inbound/jhamrick+ps1+2018-04-22 14:29:46.167901 UTC\n[SubmitApp | WARNING] Possible missing notebooks and/or extra notebooks submitted for assignment ps1:\n Expected:\n \tproblem1.ipynb: MISSING\n \tproblem2.ipynb: FOUND\n Submitted:\n \tmyproblem1.ipynb: EXTRA\n \tproblem2.ipynb: OK\n[SubmitApp | INFO] Submitted as: example_course ps1 2018-04-22 14:29:46.167901 UTC\n" ] ], [ [ "By default this assignment will still be submitted however only the \"FOUND\" notebooks (for the given assignment) can be ``autograded`` and will appear on the ``formgrade`` extension. \"EXTRA\" notebooks will not be ``autograded`` and will not appear on the ``formgrade`` extension.", "_____no_output_____" ], [ "To ensure that students cannot submit an assignment with missing notebooks (for a given assignment) the ``strict`` option, in the student's ``nbgrader_config.py`` file, can be set to ``True``:", "_____no_output_____" ] ], [ [ "%%file /tmp/student_home/nbgrader_config.py\n\nc = get_config()\nc.Exchange.root = '/tmp/exchange'\nc.Exchange.course_id = \"example_course\"\nc.ExchangeSubmit.strict = True", "Overwriting /tmp/student_home/nbgrader_config.py\n" ], [ "%%bash\nexport HOME=/tmp/student_home && cd $HOME\n\nnbgrader submit \"ps1\"", "[SubmitApp | INFO] Source: /private/tmp/student_home/ps1\n[SubmitApp | INFO] Destination: /tmp/exchange/example_course/inbound/jhamrick+ps1+2018-04-22 14:29:47.497419 UTC\n[SubmitApp | CRITICAL] Assignment ps1 not submitted. There are missing notebooks for the submission:\n Expected:\n \tproblem1.ipynb: MISSING\n \tproblem2.ipynb: FOUND\n Submitted:\n \tmyproblem1.ipynb: EXTRA\n \tproblem2.ipynb: OK\n[SubmitApp | ERROR] nbgrader submit failed\n" ] ], [ [ "## Collecting assignments", "_____no_output_____" ] ], [ [ ".. seealso::\n\n :doc:`creating_and_grading_assignments`\n Details on grading assignments after they have been collected\n\n :doc:`/command_line_tools/nbgrader-collect`\n Command line options for ``nbgrader fetch``\n\n :doc:`/command_line_tools/nbgrader-list`\n Command line options for ``nbgrader list``\n\n :doc:`philosophy`\n More details on how the nbgrader hierarchy is structured.\n\n :doc:`/configuration/config_options`\n Details on ``nbgrader_config.py``", "_____no_output_____" ] ], [ [ "First, as a reminder, here is what the instructor's `nbgrader_config.py` file looks like:", "_____no_output_____" ] ], [ [ "%%bash\n\ncat nbgrader_config.py", "\nc = get_config()\n\nc.Exchange.course_id = \"example_course\"\nc.Exchange.root = \"/tmp/exchange\"" ] ], [ [ "### From the formgrader", "_____no_output_____" ], [ "From the formgrader extension, we can collect submissions by clicking on the \"collect\" button:\n\n\n\nAs with releasing, this will display a pop-up window when the operation is complete, telling you how many submissions were collected:\n\n\n\nFrom here, you can click on the number of submissions to grade the collected submissions:\n\n", "_____no_output_____" ], [ "### From the command line", "_____no_output_____" ], [ "After students have submitted their assignments, the instructor can view what has been submitted with `nbgrader list --inbound`:", "_____no_output_____" ] ], [ [ "%%bash\n\nnbgrader list --inbound", "[ListApp | INFO] Submitted assignments:\n[ListApp | INFO] example_course jhamrick ps1 2018-04-22 14:29:40.397476 UTC\n[ListApp | INFO] example_course jhamrick ps1 2018-04-22 14:29:43.070290 UTC\n[ListApp | INFO] example_course jhamrick ps1 2018-04-22 14:29:46.167901 UTC\n" ] ], [ [ "The instructor can then collect all submitted assignments with `nbgrader collect` and passing the name of the assignment (and as with the other nbgrader commands for instructors, this must be run from the root of the course directory):", "_____no_output_____" ] ], [ [ "%%bash\n\nnbgrader collect \"ps1\"", "[CollectApp | INFO] Processing 1 submissions of 'ps1' for course 'example_course'\n[CollectApp | INFO] Collecting submission: jhamrick ps1\n" ] ], [ [ "This will copy the student submissions to the `submitted` folder in a way that is automatically compatible with `nbgrader autograde`:", "_____no_output_____" ] ], [ [ "%%bash\n\nls -l submitted", "total 0\ndrwxr-xr-x 3 jhamrick staff 96 May 31 2017 bitdiddle\ndrwxr-xr-x 3 jhamrick staff 96 May 31 2017 hacker\ndrwxr-xr-x 3 jhamrick staff 96 Apr 22 15:29 jhamrick\n" ] ], [ [ "Note that there should only ever be *one* instructor who runs the `nbgrader release` and `nbgrader collect` commands (and there should probably only be one instructor -- the same instructor -- who runs `nbgrader assign`, `nbgrader autograde` and the formgrader as well). However this does not mean that only one instructor can do the grading, it just means that only one instructor manages the assignment files. Other instructors can still perform grading by accessing the notebook server running the formgrader.", "_____no_output_____" ] ] ]

[ "markdown", "raw", "markdown", "code", "markdown", "code", "markdown", "raw", "markdown", "code", "markdown", "code", "markdown", "raw", "markdown", "raw", "markdown", "code", "markdown", "code", "markdown", "raw", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "raw", "markdown", "raw", "markdown", "raw", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "raw", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown" ]

[ [ "markdown", "markdown" ], [ "raw" ], [ "markdown", "markdown" ], [ "code" ], [ "markdown" ], [ "code" ], [ "markdown" ], [ "raw" ], [ "markdown", "markdown", "markdown", "markdown" ], [ "code" ], [ "markdown" ], [ "code" ], [ "markdown" ], [ "raw" ], [ "markdown" ], [ "raw" ], [ "markdown" ], [ "code" ], [ "markdown" ], [ "code" ], [ "markdown" ], [ "raw" ], [ "markdown", "markdown", "markdown", "markdown", "markdown" ], [ "code" ], [ "markdown" ], [ "code" ], [ "markdown" ], [ "code" ], [ "markdown", "markdown" ], [ "raw" ], [ "markdown" ], [ "raw" ], [ "markdown", "markdown" ], [ "raw" ], [ "markdown", "markdown", "markdown", "markdown", "markdown" ], [ "code" ], [ "markdown" ], [ "code" ], [ "markdown" ], [ "code" ], [ "markdown" ], [ "code" ], [ "markdown" ], [ "code" ], [ "markdown" ], [ "code" ], [ "markdown", "markdown" ], [ "code" ], [ "markdown", "markdown" ], [ "code", "code" ], [ "markdown" ], [ "raw" ], [ "markdown" ], [ "code" ], [ "markdown", "markdown", "markdown", "markdown" ], [ "code" ], [ "markdown" ], [ "code" ], [ "markdown" ], [ "code" ], [ "markdown" ] ]

[ "MIT" ]

[ "MIT" ]

[ "MIT" ]

[ [ [ "import numpy as np\nimport pandas as pd\nfrom sklearn.model_selection import train_test_split\nfrom sklearn.naive_bayes import GaussianNB\nfrom sklearn.metrics import accuracy_score,confusion_matrix\nimport matplotlib.pyplot as plt", "_____no_output_____" ], [ "import pandas as pd\n#imported the dataset given to us\npid=pd.read_csv(\"C:\\\\Users\\\\Micontroller Lab N16\\\\IIT MANDI\\\\3rd Week\\\\pima-indians-diabetes.csv\",sep=',')\n\n#made a copy of the original dataset\npid1=pid.copy()\nprint(pid1)", "_____no_output_____" ], [ "pid.columns\n#created a list of all the columns present in the dataset\npid_col=list(pid.columns)\npid2=pid.copy() #made a copy of original dataset without attribute class\npid_col\npid_col1=pid_col.copy() #we do not want to bring changes in the class column so we removed it from the copied dataset\npid_col1.remove('class')\npid2.drop([\"class\"], axis = 1, inplace = True)\nprint(pid_col1)\nprint('\\n\\n')\nprint(pid2)", "_____no_output_____" ], [ "X1 = pid2 # X denotes the input functions and here class defines whether the person is ill or not\nprint(X1)\ny1 = pid['class'] #y denotes the output functions\nprint(y1)", "_____no_output_____" ], [ "from sklearn.model_selection import train_test_split #As given we are assigning 70% of data for training and 30% for testing\nX1_train, X1_test, y1_train, y1_test = train_test_split(X1, y1, train_size = 0.7,random_state = 42)\n\nprint(X1_train)\nprint(y1_train)\nprint(X1_test)\nprint(y1_test)", "_____no_output_____" ], [ "import numpy as np\nimport matplotlib.pyplot as plt\nfrom sklearn.neighbors import KNeighborsClassifier\nfrom sklearn.metrics import accuracy_score,confusion_matrix\nneighbors=[1,3,5,7,9,11,13,15,17,19,21]\ntrain_accuracy = np.empty(len(neighbors))\ntest_accuracy = np.empty(len(neighbors))\nacc=[]\n \n# Loop over K values\nfor i, k in enumerate(neighbors):\n knn = KNeighborsClassifier(n_neighbors=k)\n knn.fit(X1_train, y1_train)\n print('Predicted Outcomes for neighbours =',k,'are', knn.predict(X1_test))\n print('\\n')\n \n print('Accuracy = ',knn.score(X1_test, y1_test))\n if ((knn.score(X1_test, y1_test))>=0):\n acc.append(knn.score(X1_test, y1_test))\n print('\\n')\n \n matrix = confusion_matrix(y1_test,knn.predict(X1_test))\n print('Confusion Matrix = ',matrix)\n print('\\n\\n')\n \n \n # Compute traning and test data accuracy\n train_accuracy[i] = knn.score(X1_train, y1_train)\n test_accuracy[i] = knn.score(X1_test, y1_test)\n\nprint(acc)\nprint('\\n')\nprint('maximum accuracy is =', max(acc)*100)\n\n \n# Generate plot\nplt.plot(neighbors, test_accuracy, label = 'Testing dataset Accuracy')\nplt.plot(neighbors, train_accuracy, label = 'Training dataset Accuracy')\n\n\nplt.legend()\nplt.xlabel('n_neighbors')\nplt.ylabel('Accuracy')\nplt.show()\n\n", "_____no_output_____" ], [ "X2 = pid2 # X denotes the input functions and here class defines whether the person is ill or not\nprint(X1)\ny2 = pid['class'] #y denotes the output functions\nprint(y1)", "_____no_output_____" ], [ "from sklearn.model_selection import train_test_split #As given we are assigning 70% of data for training and 30% for testing\nX2_train, X2_test, y2_train, y2_test = train_test_split(X2, y2, train_size = 0.7,random_state = 42)\n\nprint(X2_train)\nprint(y2_train)\nprint(X2_test)\nprint(y2_test)", "_____no_output_____" ], [ "model=GaussianNB()\nmodel.fit(X2_train, y2_train)", "_____no_output_____" ], [ "y_pred=model.predict(X2_test)\ny_pred", "_____no_output_____" ], [ "accuracy=accuracy_score(y2_test, y_pred)*100\nprint('accuracy = ',accuracy,'%')\nprint('\\n')\nmatrix=confusion_matrix(y2_test,y_pred)\nprint('Confusion Matrix = ',matrix)", "_____no_output_____" ] ], [ [ "Common way of speeding up a machine learning algorithm is by using Principal Component Analysis (PCA). \nIf your learning algorithm is too slow because the input dimension is too high, then using PCA to speed it up can be a reasonable choice.", "_____no_output_____" ] ], [ [ "X3 = pid2 # X denotes the input functions and here class defines whether the person is ill or not\nprint(X1)\ny3 = pid['class'] #y denotes the output functions\nprint(y1)", "_____no_output_____" ], [ "from sklearn.model_selection import train_test_split #As given we are assigning 70% of data for training and 30% for testing\nX3_train, X3_test, y3_train, y3_test = train_test_split(X3, y3, train_size = 0.7,random_state = 42)\n\nprint(X3_train)\nprint('\\n\\n')\nprint(y3_train)\nprint('\\n\\n')\nprint(X3_test)\nprint('\\n\\n')\nprint(y3_test)", "_____no_output_____" ], [ "from sklearn.preprocessing import StandardScaler\nscaler = StandardScaler()\n\n# Fit on training set only.\nscaler.fit(X3_train)\nprint(X3_train)\n\n# Apply transform to both the training set and the test set.\nX3_train = scaler.transform(X3_train) #expecting 2D array so only X values can be passed\nX3_test = scaler.transform(X3_test)\n", "_____no_output_____" ], [ " \n\n#the code has .95 for the number of components parameter. It means that scikit-learn choose the minimum number of principal components such that 95% of the variance is retained", "_____no_output_____" ], [ " #number of components PCA choose after fitting the model ", "_____no_output_____" ], [ "from sklearn.decomposition import PCA\n# Make an instance of the Model\ndimension = [0.25,0.40,0.60,0.70,0.80,0.90,0.95,0.99]\naccu=[]\nfor z in enumerate(dimension):\n \n pca = PCA(z) \n\n pca.fit(X3_train)\n pca.n_components_ \n\n X3_train = pca.transform(X3_train)\n X3_test= pca.transform(X3_test)\n\n\n from sklearn.neighbors import KNeighborsClassifier\n from sklearn.metrics import accuracy_score,confusion_matrix\n neighbors=[1,3,5,7,9,11,13,15,17,19,21]\n train_accuracy = np.empty(len(neighbors))\n test_accuracy = np.empty(len(neighbors))\n acc=[]\n\n # Loop over K values\n for i, k in enumerate(neighbors):\n knn = KNeighborsClassifier(n_neighbors=k)\n knn.fit(X3_train, y3_train)\n print('Predicted Outcomes for neighbours =',k,'are', knn.predict(X3_test))\n print('\\n')\n\n print('Accuracy = ',knn.score(X3_test, y3_test))\n if ((knn.score(X3_test, y3_test))>=0):\n acc.append(knn.score(X3_test, y3_test)) \n\n print('\\n')\n\n matrix = confusion_matrix(y3_test,knn.predict(X3_test))\n print('Confusion Matrix = ',matrix)\n print('\\n\\n')\n\n\n # Compute traning and test data accuracy\n train_accuracy[i] = knn.score(X3_train, y3_train)\n test_accuracy[i] = knn.score(X3_test, y3_test)\n\n print(acc)\n print('\\n')\n print('maximum accuracy is =', max(acc)*100)\n\n \n# Generate plot\n#plt.plot(neighbors, test_accuracy, label = 'Testing dataset Accuracy')\n#plt.plot(neighbors, train_accuracy, label = 'Training dataset Accuracy')\n\n\n#plt.legend()\n#plt.xlabel('n_neighbors')\n#plt.ylabel('Accuracy')\n#plt.show()", "_____no_output_____" ] ] ]

[ "code", "markdown", "code" ]

[ [ "code", "code", "code", "code", "code", "code", "code", "code", "code", "code", "code" ], [ "markdown" ], [ "code", "code", "code", "code", "code", "code" ] ]

[ "MIT" ]

[ "MIT" ]

[ "MIT" ]

[ [ [ "# The Transformer Decoder: Ungraded Lab Notebook\n\nIn this notebook, you'll explore the transformer decoder and how to implement it with Trax. \n\n## Background\n\nIn the last lecture notebook, you saw how to translate the mathematics of attention into NumPy code. Here, you'll see how multi-head causal attention fits into a GPT-2 transformer decoder, and how to build one with Trax layers. In the assignment notebook, you'll implement causal attention from scratch, but here, you'll exploit the handy-dandy `tl.CausalAttention()` layer.\n\nThe schematic below illustrates the components and flow of a transformer decoder. Note that while the algorithm diagram flows from the bottom to the top, the overview and subsequent Trax layer codes are top-down.\n\n<img src=\"transformer_decoder_lnb_figs/C4_W2_L6_transformer-decoder_S01_transformer-decoder.png\" width=\"1000\"/>", "_____no_output_____" ], [ "## Imports", "_____no_output_____" ] ], [ [ "import sys\nimport os\n\nimport time\nimport numpy as np\nimport gin\n\nimport textwrap\nwrapper = textwrap.TextWrapper(width=70)\n\nimport trax\nfrom trax import layers as tl\nfrom trax.fastmath import numpy as jnp\n\n# to print the entire np array\nnp.set_printoptions(threshold=sys.maxsize)", "INFO:tensorflow:tokens_length=568 inputs_length=512 targets_length=114 noise_density=0.15 mean_noise_span_length=3.0 \n" ] ], [ [ "## Sentence gets embedded, add positional encoding\nEmbed the words, then create vectors representing each word's position in each sentence $\\in \\{ 0, 1, 2, \\ldots , K\\}$ = `range(max_len)`, where `max_len` = $K+1$)", "_____no_output_____" ] ], [ [ "def PositionalEncoder(vocab_size, d_model, dropout, max_len, mode):\n \"\"\"Returns a list of layers that: \n 1. takes a block of text as input, \n 2. embeds the words in that text, and \n 3. adds positional encoding, \n i.e. associates a number in range(max_len) with \n each word in each sentence of embedded input text \n \n The input is a list of tokenized blocks of text\n \n Args:\n vocab_size (int): vocab size.\n d_model (int): depth of embedding.\n dropout (float): dropout rate (how much to drop out).\n max_len (int): maximum symbol length for positional encoding.\n mode (str): 'train' or 'eval'.\n \"\"\"\n # Embedding inputs and positional encoder\n return [ \n # Add embedding layer of dimension (vocab_size, d_model)\n tl.Embedding(vocab_size, d_model), \n # Use dropout with rate and mode specified\n tl.Dropout(rate=dropout, mode=mode), \n # Add positional encoding layer with maximum input length and mode specified\n tl.PositionalEncoding(max_len=max_len, mode=mode)] ", "_____no_output_____" ] ], [ [ "## Multi-head causal attention\n\nThe layers and array dimensions involved in multi-head causal attention (which looks at previous words in the input text) are summarized in the figure below: \n\n<img src=\"transformer_decoder_lnb_figs/C4_W2_L5_multi-head-attention_S05_multi-head-attention-concatenation_stripped.png\" width=\"1000\"/>\n\n`tl.CausalAttention()` does all of this for you! You might be wondering, though, whether you need to pass in your input text 3 times, since for causal attention, the queries Q, keys K, and values V all come from the same source. Fortunately, `tl.CausalAttention()` handles this as well by making use of the [`tl.Branch()`](https://trax-ml.readthedocs.io/en/latest/trax.layers.html#module-trax.layers.combinators) combinator layer. In general, each branch within a `tl.Branch()` layer performs parallel operations on copies of the layer's inputs. For causal attention, each branch (representing Q, K, and V) applies a linear transformation (i.e. a dense layer without a subsequent activation) to its copy of the input, then splits that result into heads. You can see the syntax for this in the screenshot from the `trax.layers.attention.py` [source code](https://github.com/google/trax/blob/master/trax/layers/attention.py) below: \n\n<img src=\"transformer_decoder_lnb_figs/use-of-tl-Branch-in-tl-CausalAttention.png\" width=\"500\"/>", "_____no_output_____" ], [ "## Feed-forward layer \n* Typically ends with a ReLU activation, but we'll leave open the possibility of a different activation\n* Most of the parameters are here", "_____no_output_____" ] ], [ [ "def FeedForward(d_model, d_ff, dropout, mode, ff_activation):\n \"\"\"Returns a list of layers that implements a feed-forward block.\n\n The input is an activation tensor.\n\n Args:\n d_model (int): depth of embedding.\n d_ff (int): depth of feed-forward layer.\n dropout (float): dropout rate (how much to drop out).\n mode (str): 'train' or 'eval'.\n ff_activation (function): the non-linearity in feed-forward layer.\n\n Returns:\n list: list of trax.layers.combinators.Serial that maps an activation tensor to an activation tensor.\n \"\"\"\n \n # Create feed-forward block (list) with two dense layers with dropout and input normalized\n return [ \n # Normalize layer inputs\n tl.LayerNorm(), \n # Add first feed forward (dense) layer (don't forget to set the correct value for n_units)\n tl.Dense(d_ff), \n # Add activation function passed in as a parameter (you need to call it!)\n ff_activation(), # Generally ReLU\n # Add dropout with rate and mode specified (i.e., don't use dropout during evaluation)\n tl.Dropout(rate=dropout, mode=mode), \n # Add second feed forward layer (don't forget to set the correct value for n_units)\n tl.Dense(d_model), \n # Add dropout with rate and mode specified (i.e., don't use dropout during evaluation)\n tl.Dropout(rate=dropout, mode=mode) \n ]", "_____no_output_____" ] ], [ [ "## Decoder block\nHere, we return a list containing two residual blocks. The first wraps around the causal attention layer, whose inputs are normalized and to which we apply dropout regulation. The second wraps around the feed-forward layer. You may notice that the second call to `tl.Residual()` doesn't call a normalization layer before calling the feed-forward layer. This is because the normalization layer is included in the feed-forward layer.", "_____no_output_____" ] ], [ [ "def DecoderBlock(d_model, d_ff, n_heads,\n dropout, mode, ff_activation):\n \"\"\"Returns a list of layers that implements a Transformer decoder block.\n\n The input is an activation tensor.\n\n Args:\n d_model (int): depth of embedding.\n d_ff (int): depth of feed-forward layer.\n n_heads (int): number of attention heads.\n dropout (float): dropout rate (how much to drop out).\n mode (str): 'train' or 'eval'.\n ff_activation (function): the non-linearity in feed-forward layer.\n\n Returns:\n list: list of trax.layers.combinators.Serial that maps an activation tensor to an activation tensor.\n \"\"\"\n \n # Add list of two Residual blocks: the attention with normalization and dropout and feed-forward blocks\n return [\n tl.Residual(\n # Normalize layer input\n tl.LayerNorm(), \n # Add causal attention \n tl.CausalAttention(d_feature, n_heads=n_heads, dropout=dropout, mode=mode) \n ),\n tl.Residual(\n # Add feed-forward block\n # We don't need to normalize the layer inputs here. The feed-forward block takes care of that for us.\n FeedForward(d_model, d_ff, dropout, mode, ff_activation)\n ),\n ]", "_____no_output_____" ] ], [ [ "## The transformer decoder: putting it all together\n## A.k.a. repeat N times, dense layer and softmax for output", "_____no_output_____" ] ], [ [ "def TransformerLM(vocab_size=33300,\n d_model=512,\n d_ff=2048,\n n_layers=6,\n n_heads=8,\n dropout=0.1,\n max_len=4096,\n mode='train',\n ff_activation=tl.Relu):\n \"\"\"Returns a Transformer language model.\n\n The input to the model is a tensor of tokens. (This model uses only the\n decoder part of the overall Transformer.)\n\n Args:\n vocab_size (int): vocab size.\n d_model (int): depth of embedding.\n d_ff (int): depth of feed-forward layer.\n n_layers (int): number of decoder layers.\n n_heads (int): number of attention heads.\n dropout (float): dropout rate (how much to drop out).\n max_len (int): maximum symbol length for positional encoding.\n mode (str): 'train', 'eval' or 'predict', predict mode is for fast inference.\n ff_activation (function): the non-linearity in feed-forward layer.\n\n Returns:\n trax.layers.combinators.Serial: A Transformer language model as a layer that maps from a tensor of tokens\n to activations over a vocab set.\n \"\"\"\n \n # Create stack (list) of decoder blocks with n_layers with necessary parameters\n decoder_blocks = [ \n DecoderBlock(d_model, d_ff, n_heads, dropout, mode, ff_activation) for _ in range(n_layers)] \n\n # Create the complete model as written in the figure\n return tl.Serial(\n # Use teacher forcing (feed output of previous step to current step)\n tl.ShiftRight(mode=mode), \n # Add embedding inputs and positional encoder\n PositionalEncoder(vocab_size, d_model, dropout, max_len, mode),\n # Add decoder blocks\n decoder_blocks, \n # Normalize layer\n tl.LayerNorm(), \n\n # Add dense layer of vocab_size (since need to select a word to translate to)\n # (a.k.a., logits layer. Note: activation already set by ff_activation)\n tl.Dense(vocab_size), \n # Get probabilities with Logsoftmax\n tl.LogSoftmax() \n )", "_____no_output_____" ] ], [ [ "## Concluding remarks\n\nIn this week's assignment, you'll see how to train a transformer decoder on the [cnn_dailymail](https://www.tensorflow.org/datasets/catalog/cnn_dailymail) dataset, available from TensorFlow Datasets (part of TensorFlow Data Services). Because training such a model from scratch is time-intensive, you'll use a pre-trained model to summarize documents later in the assignment. Due to time and storage concerns, we will also not train the decoder on a different summarization dataset in this lab. If you have the time and space, we encourage you to explore the other [summarization](https://www.tensorflow.org/datasets/catalog/overview#summarization) datasets at TensorFlow Datasets. Which of them might suit your purposes better than the `cnn_dailymail` dataset? Where else can you find datasets for text summarization models?", "_____no_output_____" ] ] ]

[ "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown" ]

[ [ "markdown", "markdown" ], [ "code" ], [ "markdown" ], [ "code" ], [ "markdown", "markdown" ], [ "code" ], [ "markdown" ], [ "code" ], [ "markdown" ], [ "code" ], [ "markdown" ] ]

[ "MIT" ]

[ "MIT" ]

[ "MIT" ]

[ [ [ "# Summary statistics in VCF format\n\nmodified from the create_vcf of mrcieu/gwasvcf package to transform the mash output matrixs from the rds format into a vcf file, with a effect size = to the coef and the se = 1, named as EF:SE.\n\nInput:\na collection of gene-level rds file, each file is a matrix of mash output, with colnames = studies, rownames = snps, snps shall be in the form of chr:pos_alt_ref,\nA list of aforementioned MASH output\n\noutput:\nA collection of gene-level vcf output:vcf file and corresponding index\na list of aforementioned vcf\n\nRequired R packages:\n dplyr\n readr\n VariantAnnotation\n \n\nThe output format of this workflow is following this specification https://github.com/MRCIEU/gwas-vcf-specification, with each study stands for a column", "_____no_output_____" ] ], [ [ "[global]\nimport glob\n# single column file each line is the data filename\nparameter: analysis_units = path\n# Path to data directory\nparameter: data_dir = \"/\"\n# data file suffix\nparameter: data_suffix = \"\"\n# Path to work directory where output locates\nparameter: wd = path(\"./output\")\n# An identifier for your run of analysis\nparameter: name = \"\"\n\nregions = [x.replace(\"\\\"\", \"\" ).strip().split() for x in open(analysis_units).readlines() if x.strip() and not x.strip().startswith('#')]\ngenes = regions\n# Containers that contains the necessary packages\nparameter: container = 'gaow/twas'\n", "_____no_output_____" ], [ "[rds_to_vcf_1]\ninput: genes, group_by = 1\noutput: vcf = f'{wd:a}/mash_vcf/{_input:bn}.vcf.bgz'\ntask: trunk_workers = 1, walltime = '1h', trunk_size = 1, mem = '10G', cores = 1, tags = f'{_output:bn}'\nR: expand = '$[ ]', stdout = f\"{_output[0]:nn}.stdout\", stderr = f\"{_output[0]:nn}.stderr\"\n library(\"dplyr\")\n library(\"stringr\")\n library(\"readr\")\n library(\"purrr\")\n ## Define a wrapper, modified from the gwasvcf packages, to create the vcf of needed.\n \n create_vcf = function (chrom, pos, nea, ea, snp = NULL, ea_af = NULL, effect = NULL, \n se = NULL, pval = NULL, n = NULL, ncase = NULL, name = NULL) \n {\n stopifnot(length(chrom) == length(pos))\n if (is.null(snp)) {\n snp <- paste0(chrom, \":\", pos)\n }\n snp <- paste0(chrom, \":\", pos)\n nsnp <- length(chrom)\n gen <- list()\n ## Setupt data content for each sample column\n if (!is.null(ea_af)) \n gen[[\"AF\"]] <- matrix(ea_af, nsnp)\n if (!is.null(effect)) \n gen[[\"ES\"]] <- matrix(effect, nsnp)\n if (!is.null(se)) \n gen[[\"SE\"]] <- matrix(se, nsnp)\n if (!is.null(pval)) \n gen[[\"LP\"]] <- matrix(-log10(pval), nsnp)\n if (!is.null(n)) \n gen[[\"SS\"]] <- matrix(n, nsnp)\n if (!is.null(ncase)) \n gen[[\"NC\"]] <- matrix(ncase, nsnp)\n gen <- S4Vectors::SimpleList(gen)\n \n ## Setup snps info for the fix columns\n gr <- GenomicRanges::GRanges(chrom, IRanges::IRanges(start = pos, \n end = pos + pmax(nchar(nea), nchar(ea)) - 1, names = snp))\n ## Setup meta informations\n coldata <- S4Vectors::DataFrame(Studies = name, row.names = name)\n hdr <- VariantAnnotation::VCFHeader(header = IRanges::DataFrameList(fileformat = S4Vectors::DataFrame(Value = \"VCFv4.2\", \n row.names = \"fileformat\")), sample = name)\n VariantAnnotation::geno(hdr) <- S4Vectors::DataFrame(Number = c(\"A\", \n \"A\", \"A\", \"A\", \"A\", \"A\"), Type = c(\"Float\", \"Float\", \n \"Float\", \"Float\", \"Float\", \"Float\"), Description = c(\"Effect size estimate relative to the alternative allele\", \n \"Standard error of effect size estimate\", \"-log10 p-value for effect estimate\", \n \"Alternate allele frequency in the association study\", \n \"Sample size used to estimate genetic effect\", \"Number of cases used to estimate genetic effect\"), \n row.names = c(\"ES\", \"SE\", \"LP\", \"AF\", \"SS\", \"NC\"))\n ## Save only the meta information in the sample columns \n VariantAnnotation::geno(hdr) <- subset(VariantAnnotation::geno(hdr), \n rownames(VariantAnnotation::geno(hdr)) %in% names(gen))\n ## Save VCF values\n vcf <- VariantAnnotation::VCF(rowRanges = gr, colData = coldata, \n exptData = list(header = hdr), geno = gen)\n VariantAnnotation::alt(vcf) <- Biostrings::DNAStringSetList(as.list(ea))\n VariantAnnotation::ref(vcf) <- Biostrings::DNAStringSet(nea)\n ## Write fixed values\n VariantAnnotation::fixed(vcf)$FILTER <- \"PASS\"\n return(sort(vcf))\n }\n \n input = readRDS($[_input:r])\n input_effect = input$PosteriorMean\n if(is.null(input$PosteriorSD)){\n input$PosteriorSD = matrix(1,nrow = nrow(input_effect),ncol = ncol(input_effect) )\n }\n input_se = input$PosteriorSD\n df = tibble(snps = input$snps)\n df = df%>%mutate( chr = map_dbl(snps,~str_remove(read.table(text = .x,sep = \":\",as.is = T)$V1, \"chr\")%>%as.numeric),\n pos_alt_ref = map_chr(snps,~read.table(text = .x,sep = \":\",as.is = TRUE)$V2),\n pos = map_dbl(pos_alt_ref,~read.table(text = .x,sep = \"_\",as.is = TRUE)$V1),\n alt = map_chr(pos_alt_ref,~read.table(text = .x,sep = \"_\",as.is = TRUE, colClass = \"character\")$V2),\n ref = map_chr(pos_alt_ref,~read.table(text = .x,sep = \"_\",as.is = TRUE, colClass = \"character\")$V3))\n \n \n vcf = create_vcf(\n chrom = df$chr,\n pos = df$pos,\n ea = df$alt,\n nea = df$ref,\n effect = input_effect ,\n se = input_se,\n name = colnames(input_effect))\n \n VariantAnnotation::writeVcf(vcf,$[_output:nr],index = TRUE)\n ", "_____no_output_____" ], [ "[rds_to_vcf_2]\ninput: group_by = \"all\"\noutput: vcf_list = f'{_input[0]:d}/vcf_output_list.txt'\nbash: expand = '${ }', stdout = f\"{_output[0]:nn}.stdout\", stderr = f\"{_output[0]:nn}.stderr\"\n cd ${_input[0]:d}\n ls *.vcf.bgz > ${_output}", "_____no_output_____" ] ] ]

[ "markdown", "code" ]

[ [ "markdown" ], [ "code", "code", "code" ] ]

[ "BSD-3-Clause" ]

[ "BSD-3-Clause" ]

[ "BSD-3-Clause" ]

[ [ [ "Convolutional Dictionary Learning\n=================================\n\nThis example demonstrates the use of [prlcnscdl.ConvBPDNDictLearn_Consensus](http://sporco.rtfd.org/en/latest/modules/sporco.dictlrn.prlcnscdl.html#sporco.dictlrn.prlcnscdl.ConvBPDNDictLearn_Consensus) for learning a convolutional dictionary from a set of colour training images [[51]](http://sporco.rtfd.org/en/latest/zreferences.html#id54). The dictionary learning algorithm is based on the ADMM consensus dictionary update [[1]](http://sporco.rtfd.org/en/latest/zreferences.html#id44) [[26]](http://sporco.rtfd.org/en/latest/zreferences.html#id25).", "_____no_output_____" ] ], [ [ "from __future__ import print_function\nfrom builtins import input\n\nimport pyfftw # See https://github.com/pyFFTW/pyFFTW/issues/40\nimport numpy as np\n\nfrom sporco.dictlrn import prlcnscdl\nfrom sporco import util\nfrom sporco import signal\nfrom sporco import plot\nplot.config_notebook_plotting()", "_____no_output_____" ] ], [ [ "Load training images.", "_____no_output_____" ] ], [ [ "exim = util.ExampleImages(scaled=True, zoom=0.25)\nS1 = exim.image('barbara.png', idxexp=np.s_[10:522, 100:612])\nS2 = exim.image('kodim23.png', idxexp=np.s_[:, 60:572])\nS3 = exim.image('monarch.png', idxexp=np.s_[:, 160:672])\nS4 = exim.image('sail.png', idxexp=np.s_[:, 210:722])\nS5 = exim.image('tulips.png', idxexp=np.s_[:, 30:542])\nS = np.stack((S1, S2, S3, S4, S5), axis=3)", "_____no_output_____" ] ], [ [ "Highpass filter training images.", "_____no_output_____" ] ], [ [ "npd = 16\nfltlmbd = 5\nsl, sh = signal.tikhonov_filter(S, fltlmbd, npd)", "_____no_output_____" ] ], [ [ "Construct initial dictionary.", "_____no_output_____" ] ], [ [ "np.random.seed(12345)\nD0 = np.random.randn(8, 8, 3, 64)", "_____no_output_____" ] ], [ [ "Set regularization parameter and options for dictionary learning solver.", "_____no_output_____" ] ], [ [ "lmbda = 0.2\nopt = prlcnscdl.ConvBPDNDictLearn_Consensus.Options({'Verbose': True,\n 'MaxMainIter': 200,\n 'CBPDN': {'rho': 50.0*lmbda + 0.5},\n 'CCMOD': {'rho': 1.0, 'ZeroMean': True}})", "_____no_output_____" ] ], [ [ "Create solver object and solve.", "_____no_output_____" ] ], [ [ "d = prlcnscdl.ConvBPDNDictLearn_Consensus(D0, sh, lmbda, opt)\nD1 = d.solve()\nprint(\"ConvBPDNDictLearn_Consensus solve time: %.2fs\" %\n d.timer.elapsed('solve'))", "Itn Fnc DFid Regℓ1 \n----------------------------------\n" ] ], [ [ "Display initial and final dictionaries.", "_____no_output_____" ] ], [ [ "D1 = D1.squeeze()\nfig = plot.figure(figsize=(14, 7))\nplot.subplot(1, 2, 1)\nplot.imview(util.tiledict(D0), title='D0', fig=fig)\nplot.subplot(1, 2, 2)\nplot.imview(util.tiledict(D1), title='D1', fig=fig)\nfig.show()", "_____no_output_____" ] ], [ [ "Get iterations statistics from solver object and plot functional value", "_____no_output_____" ] ], [ [ "its = d.getitstat()\nplot.plot(its.ObjFun, xlbl='Iterations', ylbl='Functional')", "_____no_output_____" ] ] ]

[ "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code" ]

[ [ "markdown" ], [ "code" ], [ "markdown" ], [ "code" ], [ "markdown" ], [ "code" ], [ "markdown" ], [ "code" ], [ "markdown" ], [ "code" ], [ "markdown" ], [ "code" ], [ "markdown" ], [ "code" ], [ "markdown" ], [ "code" ] ]

[ "MIT" ]

[ "MIT" ]

[ "MIT" ]

[ [ [ "A little notebook to help visualise the official numbers for personal use. Absolutely no guarantees are made.\n\n**This is not a replacement for expert advice. Please listen to your local health authorities.**\n\nThe data is dynamically loaded from: https://github.com/CSSEGISandData/COVID-19", "_____no_output_____" ] ], [ [ "%matplotlib inline\n%config InlineBackend.figure_format ='retina'", "_____no_output_____" ], [ "import matplotlib.pyplot as plt\nimport pandas as pd\n\nfrom jhu_helpers import *", "_____no_output_____" ], [ "jhu = aggregte_jhu_by_state(*get_jhu_data())", "_____no_output_____" ], [ "#jhu.confirmed.columns.tolist() # print a list of all countries in the data set", "_____no_output_____" ], [ "# look at recent numbers from highly affected countries\nget_aggregate_top_n(jhu.confirmed)", "_____no_output_____" ], [ "# choose a random list of countries to plot\nplot_countries = [\n 'China',\n 'Italy',\n 'Singapore', \n 'US',\n 'France',\n 'Germany',\n]", "_____no_output_____" ], [ "plt.close(1)\nfig1, ax1 = plt.subplots(nrows=2, ncols=2, figsize=(10,8), sharex=True, num=1)\n\njhu.confirmed[plot_countries].plot(ax=ax1[0,0], logy=True)\nax1[0,0].set_ylabel('Confirmed')\n\nsmooth_rate_d = 1\njhu.infection_rate[plot_countries].rolling(smooth_rate_d, center=True, min_periods=1).mean().plot(ax=ax1[1,0], logy=False)\nax1[1,0].set_ylabel('Infection Rate per Infected')\n\njhu.recovered[plot_countries].plot(ax=ax1[0,1], logy=True)\nax1[0,1].set_ylabel('Recovered')\n\njhu.deaths[plot_countries].plot(ax=ax1[1,1], logy=True)\nax1[1,1].set_ylabel('Deaths')\n\nfig1.tight_layout()", "_____no_output_____" ], [ "# save the above figure\n#fig1.savefig('sars-covid-19_timeseries.png')", "_____no_output_____" ] ] ]

[ "markdown", "code" ]

[ [ "markdown" ], [ "code", "code", "code", "code", "code", "code", "code", "code" ] ]

[ "MIT" ]

[ "MIT" ]

[ "MIT" ]

[ [ [ "Question 1:", "_____no_output_____" ] ], [ [ "Create a checker board generator, which takes as inputs n and 2 elements to generate\nan n x n checkerboard with those two elements as alternating squares.\n\nExamples\n\nchecker_board(2, 7, 6) [\n [7, 6],\n [6, 7]\n]\n\nchecker_board(3, \"A\", \"B\") [\n [\"A\", \"B\", \"A\"],\n [\"B\", \"A\", \"B\"],\n [\"A\", \"B\", \"A\"]\n]\n\nchecker_board(4, \"c\", \"d\") [\n [\"c\", \"d\", \"c\", \"d\"],\n [\"d\", \"c\", \"d\", \"c\"],\n [\"c\", \"d\", \"c\", \"d\"],\n [\"d\", \"c\", \"d\", \"c\"]\n]\n\nchecker_board(4, \"c\", \"c\") \"invalid\"", "_____no_output_____" ] ], [ [ "Answer :", "_____no_output_____" ] ], [ [ "def checker_board(n,a,b):\n \n if a == b:\n return \"invalid\"\n \n board = []\n for i in range(n):\n temp = []\n for j in range(n):\n temp.append(a)\n a, b = b, a\n b, a = temp[0:2]\n board.append(temp) \n \n return board\n \n \nfor i in checker_board(2, 7, 6):print(i)\nprint()\n\nfor i in checker_board(3, \"A\", \"B\"):print(i)\nprint()\n\nfor i in checker_board(4, \"c\", \"d\"):print(i)\nprint()\n\nchecker_board(4, \"c\", \"c\")", "[7, 6]\n[6, 7]\n\n['A', 'B', 'A']\n['B', 'A', 'B']\n['A', 'B', 'A']\n\n['c', 'd', 'c', 'd']\n['d', 'c', 'd', 'c']\n['c', 'd', 'c', 'd']\n['d', 'c', 'd', 'c']\n\n" ] ], [ [ "Question 2:", "_____no_output_____" ] ], [ [ "A string is an almost-palindrome if, by changing only one character, you\ncan make it a palindrome. Create a function that returns True if a string is an \nalmost-palindrome and False otherwise.\n\nExamples\n\nalmost_palindrome(\"abcdcbg\") True\n# Transformed to \"abcdcba\" by changing \"g\" to \"a\".\n\nalmost_palindrome(\"abccia\") True\n# Transformed to \"abccba\" by changing \"i\" to \"b\".\n\nalmost_palindrome(\"abcdaaa\") False\n# Can't be transformed to a palindrome in exactly 1 turn.\n\nalmost_palindrome(\"1234312\") False", "_____no_output_____" ] ], [ [ "Answer :", "_____no_output_____" ] ], [ [ "import string\n\ndef isPalindrome(str_):\n return str_ == str_[::-1]\n\ndef almost_palindrome(str_):\n \n check = string.ascii_lowercase + \"0123456789\"\n \n for i in str_:\n for j in check:\n temp = str_.replace(i, j, 1)\n if isPalindrome(temp):\n return True\n \n return False\n\n\nprint(almost_palindrome(\"abcdcbg\"))\nprint(almost_palindrome(\"abccia\"))\nprint(almost_palindrome(\"abcdaaa\"))\nprint(almost_palindrome(\"1234312\"))", "True\nTrue\nFalse\nFalse\n" ] ], [ [ "Question 3:", "_____no_output_____" ] ], [ [ "Create a function that finds how many prime numbers there are, up to the given integer.\n\nExamples\n\nprime_numbers(10) 4\n# 2, 3, 5 and 7\nprime_numbers(20) 8\n# 2, 3, 5, 7, 11, 13, 17 and 19\nprime_numbers(30) 10\n# 2, 3, 5, 7, 11, 13, 17, 19, 23 and 29", "_____no_output_____" ] ], [ [ "Answer :", "_____no_output_____" ] ], [ [ "def isPrime(n):\n if n <= 1:\n return False\n \n for i in range(2, int(n**(1/2))+1):\n if n % i == 0:\n return False\n return True\n\n\ndef prime_numbers(n):\n if n<2:\n return 0\n return sum([isPrime(i) for i in range(2,n+1)])\n\n\nprint(prime_numbers(10))\nprint(prime_numbers(20))\nprint(prime_numbers(30))", "4\n8\n10\n" ] ], [ [ "Question 4:", "_____no_output_____" ] ], [ [ " If today was Monday, in two days, it would be Wednesday. Create a function that takes in a list of days\nas input and the number of days to increment by. \nReturn a list of days after n number of days has passed.\n\nExamples\n\nafter_n_days([\"Thursday\", \"Monday\"], 4) [\"Monday\", \"Friday\"]\nafter_n_days([\"Sunday\", \"Sunday\", \"Sunday\"], 1) [\"Monday\", \"Monday\", \"Monday\"]\nafter_n_days([\"Monday\", \"Tuesday\", \"Friday\"], 1) [\"Tuesday\", \"Wednesday\", \"Saturday\"]", "_____no_output_____" ] ], [ [ "Answer :", "_____no_output_____" ] ], [ [ "def after_n_days(lst,n):\n \n new_lst = []\n \n if n>=7:\n _, n = divmod(n,7)\n \n for i in lst:\n days = [\"Sunday\", \"Monday\", \"Tuesday\", \"Wednesday\", \"Thursday\", \"Friday\", \"Saturday\"]\n idx = days.index(i)\n days = [days[idx]]+days[idx+1::]+days[0:idx]\n new_lst.append(days[n])\n \n return new_lst\n\n\nprint(after_n_days([\"Thursday\", \"Monday\"], 4))\nprint(after_n_days([\"Sunday\", \"Sunday\", \"Sunday\"], 1))\nprint(after_n_days([\"Monday\", \"Tuesday\", \"Friday\"], 1))", "['Monday', 'Friday']\n['Monday', 'Monday', 'Monday']\n['Tuesday', 'Wednesday', 'Saturday']\n" ] ], [ [ "Question 5:", "_____no_output_____" ] ], [ [ "You are in the process of creating a chat application and want to add an\nanonymous name feature. This anonymous name feature will create an alias that \nconsists of two capitalized words beginning with the same letter as the users first name.\nCreate a function that determines if the list of users is mapped to a list of\nanonymous names correctly.\n\nExamples\n\nis_correct_aliases([\"Adrian M.\", \"Harriet S.\", \"Mandy T.\"],\n [\"Amazing Artichoke\", \"Hopeful Hedgehog\", \"Marvelous Mouse\"]) True\n\nis_correct_aliases([\"Rachel F.\", \"Pam G.\", \"Fred Z.\", \"Nancy K.\"],\n [\"Reassuring Rat\", \"Peaceful Panda\", \"Fantastic Frog\", \"Notable Nickel\"]) True\n\nis_correct_aliases([\"Beth T.\"], [\"Brandishing Mimosa\"]) False\n\n# Both words in \"Brandishing Mimosa\" should begin with a \"B\" - \"Brandishing Beaver\" would do the trick.", "_____no_output_____" ] ], [ [ "Answer :", "_____no_output_____" ] ], [ [ "def is_correct_aliases(lst1,lst2):\n \n bool_ = []\n for i, j in zip(lst1,lst2):\n temp = j.split()\n bool_.append(i[0] == temp[0][0] and i[0] == temp[1][0])\n return all(bool_)\n\nprint(is_correct_aliases([\"Adrian M.\", \"Harriet S.\", \"Mandy T.\"],\n [\"Amazing Artichoke\", \"Hopeful Hedgehog\", \"Marvelous Mouse\"]))\nprint(is_correct_aliases([\"Rachel F.\", \"Pam G.\", \"Fred Z.\", \"Nancy K.\"],\n [\"Reassuring Rat\", \"Peaceful Panda\", \"Fantastic Frog\", \"Notable Nickel\"]))\nprint(is_correct_aliases([\"Beth T.\"], [\"Brandishing Mimosa\"]))", "True\nTrue\nFalse\n" ] ] ]

[ "markdown", "raw", "markdown", "code", "markdown", "raw", "markdown", "code", "markdown", "raw", "markdown", "code", "markdown", "raw", "markdown", "code", "markdown", "raw", "markdown", "code" ]

[ [ "markdown" ], [ "raw" ], [ "markdown" ], [ "code" ], [ "markdown" ], [ "raw" ], [ "markdown" ], [ "code" ], [ "markdown" ], [ "raw" ], [ "markdown" ], [ "code" ], [ "markdown" ], [ "raw" ], [ "markdown" ], [ "code" ], [ "markdown" ], [ "raw" ], [ "markdown" ], [ "code" ] ]

[ "Apache-2.0" ]

[ "Apache-2.0" ]

[ "Apache-2.0" ]

[ [ [ "[](https://colab.research.google.com/github/JohnSnowLabs/spark-nlp-workshop/blob/master/jupyter/enterprise/healthcare/Disambiguation.ipynb)", "_____no_output_____" ] ], [ [ "import json\n\nwith open('251keys.json') as f:\n license_keys = json.load(f)\n\nlicense_keys.keys()\n", "_____no_output_____" ], [ "\n# Install java\nimport os\n\n! apt-get install -y openjdk-8-jdk-headless -qq > /dev/null\nos.environ[\"JAVA_HOME\"] = \"/usr/lib/jvm/java-8-openjdk-amd64\"\nos.environ[\"PATH\"] = os.environ[\"JAVA_HOME\"] + \"/bin:\" + os.environ[\"PATH\"]\n! java -version\n\n# Install pyspark\n! pip install --ignore-installed -q pyspark==2.4.4\n\nsecret = license_keys['secret']\nos.environ['SPARK_NLP_LICENSE'] = license_keys['SPARK_NLP_LICENSE']\nos.environ['JSL_OCR_LICENSE'] = license_keys['JSL_OCR_LICENSE']\nos.environ['AWS_ACCESS_KEY_ID']= license_keys['AWS_ACCESS_KEY_ID']\nos.environ['AWS_SECRET_ACCESS_KEY'] = license_keys['AWS_SECRET_ACCESS_KEY']\n\n! python -m pip install --upgrade spark-nlp-jsl==2.5.1rc1 --extra-index-url https://pypi.johnsnowlabs.com/$secret\n\nimport sparknlp\n\nprint (sparknlp.version())\n\nimport json\nfrom pyspark.ml import Pipeline\nfrom pyspark.sql import SparkSession\n\n\nfrom sparknlp.annotator import *\nfrom sparknlp_jsl.annotator import *\nfrom sparknlp.base import *\nimport sparknlp_jsl\n\n\n\ndef start(secret):\n builder = SparkSession.builder \\\n .appName(\"Spark NLP Licensed\") \\\n .master(\"local[*]\") \\\n .config(\"spark.driver.memory\", \"16G\") \\\n .config(\"spark.serializer\", \"org.apache.spark.serializer.KryoSerializer\") \\\n .config(\"spark.kryoserializer.buffer.max\", \"2000M\") \\\n .config(\"spark.jars.packages\", \"com.johnsnowlabs.nlp:spark-nlp_2.11:2.5.1\") \\\n .config(\"spark.jars\", \"https://pypi.johnsnowlabs.com/\"+secret+\"/spark-nlp-jsl-2.5.1rc1.jar\")\n \n return builder.getOrCreate()\n\n\nspark = start(secret) # if you want to start the session with custom params as in start function above\n# sparknlp_jsl.start(secret)", "openjdk version \"1.8.0_252\"\nOpenJDK Runtime Environment (build 1.8.0_252-8u252-b09-1~18.04-b09)\nOpenJDK 64-Bit Server VM (build 25.252-b09, mixed mode)\n\u001b[K |████████████████████████████████| 215.7MB 59kB/s \n\u001b[K |████████████████████████████████| 204kB 51.8MB/s \n\u001b[?25h Building wheel for pyspark (setup.py) ... \u001b[?25l\u001b[?25hdone\nLooking in indexes: https://pypi.org/simple, https://pypi.johnsnowlabs.com/9hk9l8ybo1\nCollecting spark-nlp-jsl==2.5.1rc1\n Downloading https://pypi.johnsnowlabs.com/9hk9l8ybo1/spark-nlp-jsl/spark_nlp_jsl-2.5.1rc1-py3-none-any.whl\nCollecting spark-nlp==2.5.1\n\u001b[?25l Downloading https://files.pythonhosted.org/packages/df/b4/db653f8080a446de8ce981b262d85c85c61de7e920930726da0d1c6b4c65/spark_nlp-2.5.1-py2.py3-none-any.whl (121kB)\n\u001b[K |████████████████████████████████| 122kB 2.8MB/s \n\u001b[?25hRequirement already satisfied, skipping upgrade: pyspark==2.4.4 in /usr/local/lib/python3.6/dist-packages (from spark-nlp-jsl==2.5.1rc1) (2.4.4)\nRequirement already satisfied, skipping upgrade: py4j==0.10.7 in /usr/local/lib/python3.6/dist-packages (from pyspark==2.4.4->spark-nlp-jsl==2.5.1rc1) (0.10.7)\nInstalling collected packages: spark-nlp, spark-nlp-jsl\nSuccessfully installed spark-nlp-2.5.1 spark-nlp-jsl-2.5.1rc1\n2.5.1\n" ], [ "# Sample data\ntext = \"The show also had a contestant named Donald Trump \" \\\n + \"who later defeated Christina Aguilera on the way to become Female Vocalist Champion in the 1989 edition of Star Search in the United States. \"\ndata = spark.createDataFrame([\n [text]]) \\\n .toDF(\"text\").cache()\n", "_____no_output_____" ], [ "", "_____no_output_____" ], [ "# Preprocessing pipeline\nda = DocumentAssembler().setInputCol(\"text\").setOutputCol(\"document\")\nsd = SentenceDetector().setInputCols(\"document\").setOutputCol(\"sentence\")\ntk = Tokenizer().setInputCols(\"sentence\").setOutputCol(\"token\")\nemb = WordEmbeddingsModel.pretrained().setOutputCol(\"embs\")\nsemb = SentenceEmbeddings().setInputCols(\"sentence\",\"embs\").setOutputCol(\"sentence_embeddings\")\nner = NerDLModel.pretrained().setInputCols(\"sentence\",\"token\",\"embs\").setOutputCol(\"ner\")\nnc = NerConverter().setInputCols(\"sentence\",\"token\",\"ner\").setOutputCol(\"ner_chunk\").setWhiteList([\"PER\"])\ndisambiguator = NerDisambiguator() \\\n .setS3KnowledgeBaseName(\"i-per\") \\\n .setInputCols(\"ner_chunk\", \"sentence_embeddings\") \\\n .setOutputCol(\"disambiguation\") \\\n .setNumFirstChars(5)\npl = Pipeline().setStages([da,sd,tk,emb,semb,ner,nc,disambiguator])\ndata = pl.fit(data).transform(data)\ndata.selectExpr(\"explode(disambiguation)\").show(10, False)", "glove_100d download started this may take some time.\nApproximate size to download 145.3 MB\n[OK!]\nner_dl download started this may take some time.\nApproximate size to download 13.6 MB\n[OK!]\n+---------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------+\n|col |\n+---------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------+\n|[disambiguation, 37, 48, http://en.wikipedia.org/?curid=4848272, http://en.wikipedia.org/?curid=31698421, http://en.wikipedia.org/?curid=55907961, [chunk -> Donald Trump, titles -> donald trump ::::: donald crump ::::: donald frump, links -> http://en.wikipedia.org/?curid=4848272 ::::: http://en.wikipedia.org/?curid=31698421 ::::: http://en.wikipedia.org/?curid=55907961, beginInText -> 37, scores -> 0.9637175040283449, 0.9555978783336097, 0.10186673596888873, categories -> Businesspeople, Politicians, Businesspeople, Businesspeople, Politicians, ids -> 4848272, 31698421, 55907961, endInText -> 48], []]|\n|[disambiguation, 69, 86, http://en.wikipedia.org/?curid=144171, http://en.wikipedia.org/?curid=6636454, [chunk -> Christina Aguilera, titles -> christina aguilera ::::: christina aguilar, links -> http://en.wikipedia.org/?curid=144171 ::::: http://en.wikipedia.org/?curid=6636454, beginInText -> 69, scores -> 0.975820790095742, 0.9726838470180229, categories -> Musicians, Singers, Actors, Businesspeople, Musicians, Singers, ids -> 144171, 6636454, endInText -> 86], []] |\n+---------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------+\n\n" ], [ "", "_____no_output_____" ] ] ]

[ "markdown", "code" ]

[ [ "markdown" ], [ "code", "code", "code", "code", "code", "code" ] ]

[ "MIT" ]

[ "MIT" ]

[ "MIT" ]

[ [ [ "# Supply chain physics\n*This notebook illustrates methods to investigate the physics of a supply chain*\n***\nAlessandro Tufano 2020", "_____no_output_____" ], [ "### Import packages", "_____no_output_____" ] ], [ [ "import numpy as np\nimport matplotlib.pyplot as plt\n", "_____no_output_____" ] ], [ [ "### Generate empirical demand and production \nWe define an yearly sample of production quantity $x$, and demand quantity $d$", "_____no_output_____" ] ], [ [ "number_of_sample = 365 #days\nmu_production = 105 #units per day\nsigma_production = 1 # units per day\n\nmu_demand = 100 #units per day\nsigma_demand = 0.3 # units per day\n\n\nx = np.random.normal(mu_production,sigma_production,number_of_sample) \n#d = np.random.normal(mu_demand,sigma_demand,number_of_sample) \nd = brownian(x0=mu_demand, n=365, dt=1, delta=sigma_demand, out=None) #demand stochastic process\n# represent demand\nplt.hist(d,color='orange')\nplt.hist(x,color='skyblue')\n\nplt.title('Production and Demand histogram')\nplt.xlabel('Daily rate')\nplt.ylabel('Frequency')\nplt.legend(['Demand','Production'])\n\nx = np.array(x)\nd = np.array(d)\n\nplt.figure()\nplt.plot(d)\nplt.title(\"Demand curve $d$\")\nplt.xlabel('Time in days')\nplt.ylabel('Numbar of parts')\n\nplt.figure()\nplt.plot(x)\nplt.title(\"Production curve $x$\")\nplt.xlabel('Time in days')\nplt.ylabel('Number of parts')", "_____no_output_____" ] ], [ [ "### Define the inventory function $q$ \nThe empirical inventory function $q$ is defined as the differende between production and demand, plus the residual inventory. \n$q_t = q_{t-1} + x_t - d_t$", "_____no_output_____" ] ], [ [ "q = [mu_production] #initial inventory with production mean value\nfor i in range(0,len(d)):\n inventory_value = q[i] + x[i] - d[i] \n if inventory_value <0 : \n inventory_value=0\n q.append(inventory_value)\n \nplt.plot(q)\nplt.xlabel('days')\nplt.ylabel('Inventory quantity $q$')\nplt.title('Inventory function $q$')\n\nq = np.array(q)", "_____no_output_____" ] ], [ [ "### Define pull and push forces (the momentum $p=\\dot{q}$) \nBy using continuous notation we obtain the derivative $\\dot{q}=p=x-d$. The derivative of the inventory represents the *momentum* of the supply chain, i.e. the speed a which the inventory values goes up (production), and down (demand). We use the term **productivity** to identify the momentum $p$. The forces changing the value of the productivity are called **movements** $\\dot{p}$.", "_____no_output_____" ] ], [ [ "p1 = [q[i]-q[i-1] for i in range(1,len(q))]\np2 = [x[i]-d[i] for i in range(1,len(d))]\nplt.plot(p1)\nplt.plot(p2)\nplt.xlabel('days')\nplt.ylabel('Value')\nplt.title('Momentum function $p$')\n\np=np.array(p)", "_____no_output_____" ] ], [ [ "### Define a linear potential $V(q)$ \nwe introduce a linear potential to describe the amount of *energy* related with a given quantity of the inventory $q$.", "_____no_output_____" ] ], [ [ "F0 = 0.1\n\n#eta = 1.2\n#lam = mu_demand\n#F0=eta*lam\n\nprint(F0)\n\nV_q = -F0*q\nV_q = V_q[0:-1]", "_____no_output_____" ] ], [ [ "### Define the energy conservation function using the Lagrangianm and the Hamiltonian \nWe use the Lagrangian to describe the energy conservation equation.\n$L(q,\\dot{q}) = H = \\frac{1}{2}\\dot{q} - V(q)$", "_____no_output_____" ] ], [ [ "H = (p**2)/2 - F0*q[0:-1]\n\nplt.plot(H)\nplt.xlabel('days')\nplt.ylabel('value')\nplt.title('Function $H$')", "_____no_output_____" ] ], [ [ "### Obtain the inventory $q$, given $H$", "_____no_output_____" ] ], [ [ "S_q = [H[i-1] + H[i] for i in range(1,len(H))]\nplt.plot(S_q)\nplt.xlabel('days')\nplt.ylabel('value')\nplt.title('Function $S[q]$')\n\n\n#compare with q\nplt.plot(q)\nplt.xlabel('days')\nplt.ylabel('Inventory quantity $q$')\nplt.title('Inventory function $q$')\n\nplt.legend(['Model inventory','Empirical inventory'])", "_____no_output_____" ] ], [ [ "# Inventory control", "_____no_output_____" ], [ "### Define the Brownian process", "_____no_output_____" ] ], [ [ "from math import sqrt\nfrom scipy.stats import norm\nimport numpy as np\n\n\ndef brownian(x0, n, dt, delta, out=None):\n \"\"\"\n Generate an instance of Brownian motion (i.e. the Wiener process):\n\n X(t) = X(0) + N(0, delta**2 * t; 0, t)\n\n where N(a,b; t0, t1) is a normally distributed random variable with mean a and\n variance b. The parameters t0 and t1 make explicit the statistical\n independence of N on different time intervals; that is, if [t0, t1) and\n [t2, t3) are disjoint intervals, then N(a, b; t0, t1) and N(a, b; t2, t3)\n are independent.\n \n Written as an iteration scheme,\n\n X(t + dt) = X(t) + N(0, delta**2 * dt; t, t+dt)\n\n\n If `x0` is an array (or array-like), each value in `x0` is treated as\n an initial condition, and the value returned is a numpy array with one\n more dimension than `x0`.\n\n Arguments\n ---------\n x0 : float or numpy array (or something that can be converted to a numpy array\n using numpy.asarray(x0)).\n The initial condition(s) (i.e. position(s)) of the Brownian motion.\n n : int\n The number of steps to take.\n dt : float\n The time step.\n delta : float\n delta determines the \"speed\" of the Brownian motion. The random variable\n of the position at time t, X(t), has a normal distribution whose mean is\n the position at time t=0 and whose variance is delta**2*t.\n out : numpy array or None\n If `out` is not None, it specifies the array in which to put the\n result. If `out` is None, a new numpy array is created and returned.\n\n Returns\n -------\n A numpy array of floats with shape `x0.shape + (n,)`.\n \n Note that the initial value `x0` is not included in the returned array.\n \"\"\"\n\n x0 = np.asarray(x0)\n\n # For each element of x0, generate a sample of n numbers from a\n # normal distribution.\n r = norm.rvs(size=x0.shape + (n,), scale=delta*sqrt(dt))\n\n # If `out` was not given, create an output array.\n if out is None:\n out = np.empty(r.shape)\n\n # This computes the Brownian motion by forming the cumulative sum of\n # the random samples. \n np.cumsum(r, axis=-1, out=out)\n\n # Add the initial condition.\n out += np.expand_dims(x0, axis=-1)\n\n return out", "_____no_output_____" ] ], [ [ "### Define the supply chain control model", "_____no_output_____" ] ], [ [ "# supply chain control model\ndef supply_chain_control_model(p,beta,eta,F0):\n #p is the productivity function defined as the defivative of q\n #beta is the diffusion coefficient, i.e. the delta of the Brownian process, the std of the demand can be used\n #eta represents the flexibility of the productio. It is the number of days to reach a target inventory\n #F0 is the potential \n \n Fr_t = brownian(x0=F0, n=365, dt=1, delta=beta, out=None) #demand stochastic process\n p_dot = F0 -eta*p + Fr_t\n return p_dot, Fr_t", "_____no_output_____" ], [ "#identify the sensitivity of the inventory control with different values of eta\nfor eta in [0.1,1,2,7,30]:\n p_dot, Fr_t = supply_chain_control_model(p=p,beta = sigma_demand,eta=eta,F0=F0)\n \n\n plt.figure()\n plt.plot(Fr_t)\n plt.plot(p)\n plt.plot(p_dot)\n plt.title(f\"Inventory control with eta={eta}\")\n plt.legend(['Demand','Productivity','Movements $\\dot{p}$'])", "_____no_output_____" ], [ "p_dot, Fr_t = supply_chain_control_model(p=p,beta = sigma_demand,eta=1,F0=0.9)\np_model = [p_dot[i-1] + p_dot[i] for i in range(1,len(p_dot))]\nq_model = [p_model[i-1] + p_model[i] for i in range(1,len(p_model))]\n\n\nplt.plot(q_model)\nplt.plot(p_model)\nplt.legend(['$q$: inventory','$p$: productivity'])\n\np_mean = np.mean(p_model)\np_std = np.std(p_model)\n\nprint(f\"Movements mean: {p_mean}, std: {p_std}\")\n\nq_mean = np.mean(q_model)\nq_std = np.std(q_model)\n\nprint(f\"Inventory mean: {q_mean}, std: {q_std}\")", "_____no_output_____" ] ] ]

[ "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code" ]

[ [ "markdown", "markdown" ], [ "code" ], [ "markdown" ], [ "code" ], [ "markdown" ], [ "code" ], [ "markdown" ], [ "code" ], [ "markdown" ], [ "code" ], [ "markdown" ], [ "code" ], [ "markdown" ], [ "code" ], [ "markdown", "markdown" ], [ "code" ], [ "markdown" ], [ "code", "code", "code" ] ]

[ "MIT" ]

[ "MIT" ]

[ "MIT" ]

[ [ [ "# Ingest Data\nOriginal data was from inside airbnb, to secure the files, they were copied to google drive", "_____no_output_____" ] ], [ [ "import os\nfrom google_drive_downloader import GoogleDriveDownloader as gdd", "_____no_output_____" ] ], [ [ "# Pull files from Google Drive\nlistings shared url: https://drive.google.com/file/d/1e8hVygvxFgJo3QgUrzgslsTzWD9-MQUO/view?usp=sharing\n\ncalendar shared url: https://drive.google.com/file/d/1VjlSWEr4vaJHdT9o2OF9N2Ga0X2b22v9/view?usp=sharing\n\nreviews shared url: https://drive.google.com/file/d/1_ojDocAs_LtcBLNxDHqH_TSBWjPz-Zme/view?usp=sharing\n\n", "_____no_output_____" ] ], [ [ "# gdd.download_file_from_google_drive(file_id='1e8hVygvxFgJo3QgUrzgslsTzWD9-MQUO',\n# dest_path='../data/gdrive/listings.csv.gz'", "_____no_output_____" ], [ "#source_dest = {'../data/gdrive/listings.csv.gz':'1e8hVygvxFgJo3QgUrzgslsTzWD9-MQUO'}\n\nsource_dest = {'../data/gdrive/listings.csv.gz':'1e8hVygvxFgJo3QgUrzgslsTzWD9-MQUO', '../data/gdrive/calendar.csv.gz':'1VjlSWEr4vaJHdT9o2OF9N2Ga0X2b22v9', '../data/gdrive/reviews.csv.gz':'1_ojDocAs_LtcBLNxDHqH_TSBWjPz-Zme'}\n\n", "_____no_output_____" ], [ "for k, v in source_dest.items():\n gdd.download_file_from_google_drive(file_id=v, dest_path=k)", "_____no_output_____" ] ] ]

[ "markdown", "code", "markdown", "code" ]

[ [ "markdown" ], [ "code" ], [ "markdown" ], [ "code", "code", "code" ] ]

[ "MIT" ]

[ "MIT" ]

[ "MIT" ]

[ [ [ "# Demo for paper \"First Order Motion Model for Image Animation\"\n\n---\n\n", "_____no_output_____" ], [ "**Clone repository**", "_____no_output_____" ] ], [ [ "!git clone https://github.com/hamdirhibi/Deep-fake", "_____no_output_____" ], [ "cd Deep-fake", "_____no_output_____" ] ], [ [ "\n\n```\n# This is formatted as code\n```\n\n**Mount your Google drive folder on Colab**", "_____no_output_____" ] ], [ [ "from google.colab import drive\ndrive.mount('/content/gdrive')", "_____no_output_____" ] ], [ [ "", "_____no_output_____" ], [ "**Add folder https://drive.google.com/drive/folders/157-wifsuylAkO1E4hBGO_QXyn22mDXET?usp=sharing to your google drive.\nAlternativelly you can use this mirror link https://drive.google.com/drive/folders/157-wifsuylAkO1E4hBGO_QXyn22mDXET?usp=sharing", "_____no_output_____" ], [ "**Load driving video and source image**", "_____no_output_____" ] ], [ [ "import imageio\nimport numpy as np\nimport matplotlib.pyplot as plt\nimport matplotlib.animation as animation\nfrom skimage.transform import resize\nfrom IPython.display import HTML\nimport warnings\nwarnings.filterwarnings(\"ignore\")\n\nsource_image = imageio.imread('/content/gdrive/My Drive/first-order-motion-model/rayen.jpg')\nreader = imageio.get_reader('/content/gdrive/My Drive/first-order-motion-model/hamdi.mp4')\n\n\n#Resize image and video to 256x256\n\nsource_image = resize(source_image, (256, 256))[..., :3]\n\nfps = reader.get_meta_data()['fps']\ndriving_video = []\ntry:\n for im in reader:\n driving_video.append(im)\nexcept RuntimeError:\n pass\nreader.close()\n\ndriving_video = [resize(frame, (256, 256))[..., :3] for frame in driving_video]\n\ndef display(source, driving, generated=None):\n fig = plt.figure(figsize=(8 + 4 * (generated is not None), 6))\n\n ims = []\n for i in range(len(driving)):\n cols = [source]\n cols.append(driving[i])\n if generated is not None:\n cols.append(generated[i])\n im = plt.imshow(np.concatenate(cols, axis=1), animated=True)\n plt.axis('off')\n ims.append([im])\n\n ani = animation.ArtistAnimation(fig, ims, interval=50, repeat_delay=1000)\n plt.close()\n return ani\n \n\nHTML(display(source_image, driving_video).to_html5_video())", "_____no_output_____" ] ], [ [ "**Create a model and load checkpoints**", "_____no_output_____" ] ], [ [ "from demo import load_checkpoints\ngenerator, kp_detector = load_checkpoints(config_path='config/vox-256.yaml', \n checkpoint_path='/content/gdrive/My Drive/first-order-motion-model/vox-cpk.pth.tar')", "_____no_output_____" ] ], [ [ "** **bold text**Perform image animation**", "_____no_output_____" ] ], [ [ "from demo import make_animation\nfrom skimage import img_as_ubyte\n\npredictions = make_animation(source_image, driving_video, generator, kp_detector, relative=True)\n\n#save resulting video\nimageio.mimsave('../generated.mp4', [img_as_ubyte(frame) for frame in predictions], fps=fps)\n#video can be downloaded from /content folder\n\nHTML(display(source_image, driving_video, predictions).to_html5_video())", "_____no_output_____" ] ], [ [ "**In the cell above we use relative keypoint displacement to animate the objects. We can use absolute coordinates instead, but in this way all the object proporions will be inherited from the driving video. For example Putin haircut will be extended to match Trump haircut.**", "_____no_output_____" ] ], [ [ "predictions = make_animation(source_image, driving_video, generator, kp_detector, relative=False, adapt_movement_scale=True)\nHTML(display(source_image, driving_video, predictions).to_html5_video())", "_____no_output_____" ] ], [ [ "## Running on your data\n\n**First we need to crop a face from both source image and video, while simple graphic editor like paint can be used for cropping from image. Cropping from video is more complicated. You can use ffpmeg for this.**", "_____no_output_____" ] ], [ [ "!ffmpeg -i /content/gdrive/My\\ Drive/first-order-motion-model/07.mkv -ss 00:08:57.50 -t 00:00:08 -filter:v \"crop=600:600:760:50\" -async 1 hinton.mp4", "_____no_output_____" ] ], [ [ "**Another posibility is to use some screen recording tool, or if you need to crop many images at ones use face detector(https://github.com/1adrianb/face-alignment) , see https://github.com/AliaksandrSiarohin/video-preprocessing for preprcessing of VoxCeleb.** ", "_____no_output_____" ] ], [ [ "source_image = imageio.imread('/content/gdrive/My Drive/first-order-motion-model/09.png')\ndriving_video = imageio.mimread('hinton.mp4', memtest=False)\n\n\n#Resize image and video to 256x256\n\nsource_image = resize(source_image, (256, 256))[..., :3]\ndriving_video = [resize(frame, (256, 256))[..., :3] for frame in driving_video]\n\npredictions = make_animation(source_image, driving_video, generator, kp_detector, relative=True,\n adapt_movement_scale=True)\n\nHTML(display(source_image, driving_video, predictions).to_html5_video())", "_____no_output_____" ] ] ]

[ "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code" ]

[ [ "markdown", "markdown" ], [ "code", "code" ], [ "markdown" ], [ "code" ], [ "markdown", "markdown", "markdown" ], [ "code" ], [ "markdown" ], [ "code" ], [ "markdown" ], [ "code" ], [ "markdown" ], [ "code" ], [ "markdown" ], [ "code" ], [ "markdown" ], [ "code" ] ]

[ "MIT" ]

[ "MIT" ]

[ "MIT" ]

[ [ [ "### Realizando limpeza dos dados\n##### Por Adriano Santos\n\n###### Dentre as atividades que um cientista de dados deve realizar, o processo de limpeza e tratamento é uma das mais importantes.\n\nNesta aula aprenderemos a:\n\n* Remover informações de um DataFrame;\n", "_____no_output_____" ] ], [ [ "# Carregando módulos \nimport pandas as pd", "_____no_output_____" ], [ "# Importando os dados para manipulação\ndf = pd.read_csv('../Dados/WHO.csv', delimiter=',')\nprint (df.head())", " Country Region Population Under15 Over60 \\\n0 Afghanistan Eastern Mediterranean 29825 47.42 3.82 \n1 Albania Europe 3162 21.33 14.93 \n2 Algeria Africa 38482 27.42 7.17 \n3 Andorra Europe 78 15.20 22.86 \n4 Angola Africa 20821 47.58 3.84 \n\n FertilityRate LifeExpectancy ChildMortality CellularSubscribers \\\n0 5.40 60 98.5 54.26 \n1 1.75 74 16.7 96.39 \n2 2.83 73 20.0 98.99 \n3 NaN 82 3.2 75.49 \n4 6.10 51 163.5 48.38 \n\n LiteracyRate GNI PrimarySchoolEnrollmentMale \\\n0 NaN 1140.0 NaN \n1 NaN 8820.0 NaN \n2 NaN 8310.0 98.2 \n3 NaN NaN 78.4 \n4 70.1 5230.0 93.1 \n\n PrimarySchoolEnrollmentFemale \n0 NaN \n1 NaN \n2 96.4 \n3 79.4 \n4 78.2 \n" ] ], [ [ "##### Verificando a existência de dados missing (NaN)", "_____no_output_____" ] ], [ [ "# O any() possibibilitará saber, coluna a coluna, se qualquer um dos valores é inexistente.\ndf.isnull().any()\n", "_____no_output_____" ], [ "# Possibilitirá se existe alguma coluna em branco.\nprint (df.isnull().all())\nprint ('Número de registros:', df.shape)", "Country False\nRegion False\nPopulation False\nUnder15 False\nOver60 False\nFertilityRate False\nLifeExpectancy False\nChildMortality False\nCellularSubscribers False\nLiteracyRate False\nGNI False\nPrimarySchoolEnrollmentMale False\nPrimarySchoolEnrollmentFemale False\ndtype: bool\nNúmero de registros: (194, 13)\n" ], [ "# O comando dropna() remove do DataFrame qualquer linha que tenha pelo menos um NaN.\ndf.dropna(inplace=True) # thresh=2 -> se tiver mais de dois NaN; how='all' se tiver uma linha completa em NaN\nprint ('Número de registros:', df['Country'].count())\ndf.isnull().any()", "Número de registros: 50\n" ], [ "# Carrega os dados novamente para iniciarmos outras formas de ajustes de dados.\ndf = pd.read_csv('../Dados/WHO.csv', delimiter=',')\nprint (df.shape)", "(194, 13)\n" ], [ "# Caso você queira preencher os valores inexistentes, você deve usar a função fillna()\ndf.isnull().any()\n# Iremos preencher os valores NaN da coluna \nprint('Registros:', df['FertilityRate'].count())\ndf['FertilityRate'].head()", "Registros: 183\n" ], [ "# Preenchendo os valores com os valores da média\ndf['FertilityRate'].fillna(df['FertilityRate'].mean(),inplace=True)\nprint('Número de registros:', df['FertilityRate'].count())\ndf['FertilityRate'].head()\n", "Número de registros: 194\n" ] ], [ [ "###### Comandos para remoção de coluna", "_____no_output_____" ] ], [ [ "# Para remover, faça:\ndf.drop('CellularSubscribers', axis=1, inplace=True) # axis 1 = coluna; axis 0 = linha.\nprint (df.columns)\n", "Index(['Country', 'Region', 'Population', 'Under15', 'Over60', 'FertilityRate',\n 'LifeExpectancy', 'ChildMortality', 'LiteracyRate', 'GNI',\n 'PrimarySchoolEnrollmentMale', 'PrimarySchoolEnrollmentFemale'],\n dtype='object')\n" ], [ "# Avaliando se existe duplicata \nprint(df.duplicated('Region').head())", "0 False\n1 False\n2 False\n3 True\n4 True\ndtype: bool\n" ] ] ]

[ "markdown", "code", "markdown", "code", "markdown", "code" ]

[ [ "markdown" ], [ "code", "code" ], [ "markdown" ], [ "code", "code", "code", "code", "code", "code" ], [ "markdown" ], [ "code", "code" ] ]

[ "MIT" ]

[ "MIT" ]

[ "MIT" ]

[ [ [ "import numpy as np\nimport pandas as pd\nimport matplotlib.pyplot as plt\nfrom scipy.signal import find_peaks", "_____no_output_____" ], [ "# load the appliances energy prediction data set\n\ndata = pd.read_csv('energydata_complete.csv')\n\ndata.head()", "_____no_output_____" ], [ "# parse as datetime data type\n\ndata['date'] = pd.to_datetime(data['date'])", "_____no_output_____" ], [ "# determine time between transactions, in this case, energy records\n\ndata['time_since_previous'] = data['date'].diff()\n\ndata['time_since_previous'] = data['time_since_previous']/np.timedelta64(1,'m')\n\ndata[['date', 'time_since_previous']].head(10)", "_____no_output_____" ], [ "# extract day and month from datetime variable\n\ndata[['day', 'month']] = pd.DataFrame([(x.day, x.month) for x in data['date']])\n\ndata.head()", "_____no_output_____" ], [ "# make the datetime variable the index of the series\n\ndata.index = data['date']", "_____no_output_____" ], [ "# Plot mean energy consumption by appliances per day\n# we are going to work with this data\n\ndata.groupby(['month', 'day'])['Appliances'].mean().plot()", "_____no_output_____" ], [ "# Plot mean energy consumption by lights per day\n# we are going to work with this data\n\ndata.groupby(['month', 'day'])['lights'].mean().plot()", "_____no_output_____" ], [ "# create pandas series with the mean energy consumption per day\n\n# electricity consumption by appliances per day\nelec_pday = data.groupby(['month', 'day'])['Appliances'].mean()\n\n# light energy consumption per day\nlight_pday = data.groupby(['month', 'day'])['lights'].mean()", "_____no_output_____" ], [ "# find the peaks, that is, the local maxima\n\npeaks, _ = find_peaks(elec_pday.values, height=60)\n\npeaks", "_____no_output_____" ], [ "# compare the shape of the time series with that of the selected\n# local maxima series\n\nelec_pday.shape, elec_pday[peaks].shape", "_____no_output_____" ], [ "# select only the values of the series with the local maxima\n\nelec_pday[peaks]", "_____no_output_____" ], [ "# capture the series with local maxima in a pandas dataframe\n# then reset index so that the month and day become part of the columns\n\n# finally, we need to add the year, to be able to reconstitute the date\n# from the existing time columns\n\ntmp = pd.DataFrame(elec_pday[peaks]).reset_index(drop=False)\ntmp['year'] = 2016\ntmp.head()", "_____no_output_____" ], [ "# reconstitute the datetime variable\n\ntmp['date'] = pd.to_datetime(tmp[['year', 'month', 'day']])\n\ntmp.head()", "_____no_output_____" ], [ "# calculate the distance, in days, between the local maxima\n# we do this utilizing the dataframe with only the local maxima\n\ntmp['peak_distance'] = tmp['date'].diff()\n\ntmp['peak_distance'] = tmp['peak_distance'].dt.days\n\ntmp.head()", "_____no_output_____" ], [ "# now we put all the steps together in a function\n\n# not in book, but useful information for readers, to\n# automate the calculation of peak distances across variables\n\ndef time_between_peaks(ser):\n\n # find local maxima\n peaks, _ = find_peaks(ser.values)\n\n # select the series values with local maxima only\n # transform the series into a dataframe with the month\n # and day index as columns\n tmp = pd.DataFrame(ser[peaks]).reset_index(drop=False)\n\n # add year to reconstitute date\n tmp['year'] = 2016\n\n # reconstitute date\n tmp['date'] = pd.to_datetime(tmp[['year', 'month', 'day']])\n\n # calculate difference in days between local maxima\n tmp['peak_distance'] = tmp['date'].diff()\n tmp['peak_distance'] = tmp['peak_distance'].dt.days\n\n # return difference in days between local maxima\n # that is a pandas series\n return tmp['peak_distance']", "_____no_output_____" ], [ "# return a series with a difference in days respect to the\n# previous local maxima for the time series with the\n# mean daily energy consumption by lights\n\ndistances = time_between_peaks(light_pday)\n\n# display first 10 values of the series\ndistances[0:10]", "_____no_output_____" ] ], [ [ "## To determine distance between local maxima and minima\n\nWe need to calculate both, and then concatenate the arrays, and use that to select the data", "_____no_output_____" ] ], [ [ "# determine the days of minimum electricity consumption \n# throughout the 5 months, that is the local minima\n\n# we use peak values but we turn the series upside down with the\n# reciprocal function\n\nvalleys, _ = find_peaks(1 / elec_pday.values, height=(-np.Inf, 1/60))\nvalleys", "_____no_output_____" ], [ "# compare the number of observations in the entire series\n# vs the number of local maxima, vs the number of local minima\n\nelec_pday.shape, elec_pday[peaks].shape, elec_pday[valleys].shape", "_____no_output_____" ], [ "# concatenate the indices that contain the local minima and maxima\n# and then sort its values\n\npeaksandvalleys = np.concatenate([peaks, valleys])\npeaksandvalleys.sort()\npeaksandvalleys", "_____no_output_____" ], [ "# now we use this index to select the data\n\nelec_pday[peaksandvalleys].shape", "_____no_output_____" ] ], [ [ "To determine the time elapsed between local maxima and minima, we need create a dataframe with those values executing:\n\n tmp = pd.DataFrame(elec_pday[peaksandvalleys]).reset_index(drop=False)\n\nand then, 1) add the year, 2) reconstitute the date, and 3) calculate the time between the local maxima and minima, as we have done in previous cells.", "_____no_output_____" ], [ "## There is more\n\nWe can determine the mean difference between events for various customers or entities.\n\nLet's work with the mock customer transactions data set as example.", "_____no_output_____" ] ], [ [ "import featuretools as ft\n\n# load data set from feature tools\ndata_dict = ft.demo.load_mock_customer()\n\ndata = data_dict[\"transactions\"].merge(\n data_dict[\"sessions\"]).merge(data_dict[\"customers\"])\n\ncols = ['customer_id',\n 'transaction_id',\n 'transaction_time',\n 'amount',\n ]\n\ndata = data[cols]\n\ndata.head()", "_____no_output_____" ], [ "# Let's first calculate the time since previous transaction for each transaction\n\n# sort the data by transaction date and time\ndata.sort_values(by=['transaction_time'], ascending=True, inplace=True)\n\n# calculate time since previous transaction in hours\ndata['time_since_previous'] = data['transaction_time'].diff()\ndata['time_since_previous'] = data['time_since_previous']/np.timedelta64(1,'h')", "_____no_output_____" ], [ "# calculate mean time between transactions per customer\n# all transactions occur every 1h and 5 min in the toy data set\n# so the result is a bit boring, but you get the yiest\n\ntmp = data.groupby('customer_id')['time_since_previous'].mean()\ntmp", "_____no_output_____" ], [ "# Now, let's calculate the time between local extrema\n\n# extract the hour of the transaction\n\ndata['hr'] = data['transaction_time'].dt.hour\n\ndata.head()", "_____no_output_____" ], [ "# now let's plot the local maxima for the time series with \n# the mean amount spent per hour ==>\n\n# one plot per customer\n\n# this code is intended to get the reader familiar with the data\n# and therefore facilitate the understanding of the recipe code\n\n\ndef find_and_plot_peaks(x, customer):\n\n # find local maxima and minima\n peaks, _ = find_peaks(x)\n valleys, _ = find_peaks(1/x)\n \n # plot the peaks and valleys\n plt.figure(figsize=(4,3))\n plt.plot(x)\n plt.plot(peaks, x[peaks], \"x\")\n plt.plot(valleys, x[valleys], \"x\", color='red')\n plt.title('Customer number {}'.format(customer))\n plt.show()", "_____no_output_____" ], [ "# make a plot per customer, with local minima and\n# maxima of amount spent per hr\n\nfor customer in data['customer_id'].unique():\n tmp = data[data['customer_id']==customer]\n tmp = tmp.groupby('hr')['amount'].mean()\n tmp.reset_index(drop=True, inplace=True)\n find_and_plot_peaks(tmp, customer)\n", "_____no_output_____" ], [ "# create a series of functions to find the local maxima and minima\n# put the arrays together, and then slice the original series into\n# those values, to finally calculate the time elapsed between them\n\n# these functions operate at a pandas series level\n# x is a pandas series\n\ndef find_no_peaks(x):\n # finds number of local maxima\n peaks, _ = find_peaks(x)\n return peaks\n\ndef find_no_valleys(x):\n # finds number of local minima\n valleys, _ = find_peaks(1/x)\n return valleys\n\ndef concatenate_pav(x):\n # concatenates the indeces of the peaks and valleys\n ids = np.concatenate([find_no_peaks(x), find_no_valleys(x)])\n ids.sort()\n return ids\n\ndef slice_and_measure(x):\n # selects the points with peaks and valleys in the series\n # and determines the hr difference between them.\n # finally, returns the mean distance between\n # all local maxima and minima\n ids = concatenate_pav(x)\n tmp = pd.DataFrame(x.iloc[ids]).reset_index(drop=False)\n t = tmp['hr'].diff()\n return t.mean(skipna=True)", "_____no_output_____" ], [ "# this pandas series, df, is the argument we need for the\n# precedent functions\n\n# we use the data of customer 3 as an example to test each\n# individual function\n\ndf = data[data['customer_id'] == 3]\ndf = df.groupby('hr')['amount'].mean()\ndf", "_____no_output_____" ], [ "# test function that finds number of peaks\npeaks = find_no_peaks(df)\npeaks", "_____no_output_____" ], [ "# test function that finds number of valleys\nvalleys = find_no_valleys(df)\nvalleys", "_____no_output_____" ], [ "# test concatenate function\n\nids = concatenate_pav(df)\nids", "_____no_output_____" ], [ "# test result of concatenate_pav when applied to the\n# entire dataset: that would be the indeces with max and min\n# transaction amount per customer\n\ndata.groupby(['customer_id', 'hr'])['amount'].mean().groupby('customer_id').apply(concatenate_pav)", "_____no_output_____" ], [ "# step by step the inner code of the function slide_and_measure()\n\ntmp = pd.DataFrame(df.iloc[ids]).reset_index(drop=False)\nt = tmp['hr'].diff()\nt", "_____no_output_____" ], [ "# output of slide_and_measure()\n\nt.mean(skipna=True)", "_____no_output_____" ], [ "# test slide_and_measure() on 1 customer\nslice_and_measure(df)", "_____no_output_____" ], [ "# apply slide_and_measure() to the entire data set\n\ndata.groupby(['customer_id', 'hr'])['amount'].mean().groupby(\n 'customer_id').apply(slice_and_measure)", "_____no_output_____" ] ], [ [ "Compare the distances returned by our function with the peak distances in the plots in cell **23**.\n\nCustomer 2 shows np.nan, because it only contains 1 local maxima, so it is not possible to calculate distances.", "_____no_output_____" ] ] ]

[ "code", "markdown", "code", "markdown", "code", "markdown" ]

[ [ "code", "code", "code", "code", "code", "code", "code", "code", "code", "code", "code", "code", "code", "code", "code", "code", "code" ], [ "markdown" ], [ "code", "code", "code", "code" ], [ "markdown", "markdown" ], [ "code", "code", "code", "code", "code", "code", "code", "code", "code", "code", "code", "code", "code", "code", "code", "code" ], [ "markdown" ] ]

[ "Apache-2.0" ]

[ "Apache-2.0" ]

[ "Apache-2.0" ]

[ [ [ "<h1>**Table of Contents**<span class=\"tocSkip\"></span></h1>\n<div class=\"toc\"><ul class=\"toc-item\"><li><span><a href=\"#Feature-engineering---quantifying-access-to-facilities---batch-mode\" data-toc-modified-id=\"Feature-engineering---quantifying-access-to-facilities---batch-mode-1\">Feature engineering - quantifying access to facilities - batch mode</a></span><ul class=\"toc-item\"><li><span><a href=\"#Read-the-shortlisted-properties\" data-toc-modified-id=\"Read-the-shortlisted-properties-1.1\">Read the shortlisted properties</a></span></li><li><span><a href=\"#Loop-through-each-property-and-build-the-neighborhood-facility-table\" data-toc-modified-id=\"Loop-through-each-property-and-build-the-neighborhood-facility-table-1.2\">Loop through each property and build the neighborhood facility table</a></span></li><li><span><a href=\"#Feature-engineer-with-access-to-amenities\" data-toc-modified-id=\"Feature-engineer-with-access-to-amenities-1.3\">Feature engineer with access to amenities</a></span><ul class=\"toc-item\"><li><span><a href=\"#Plot-the-distribution-of-facility-access\" data-toc-modified-id=\"Plot-the-distribution-of-facility-access-1.3.1\">Plot the distribution of facility access</a></span></li><li><span><a href=\"#Store-to-disk\" data-toc-modified-id=\"Store-to-disk-1.3.2\">Store to disk</a></span></li></ul></li></ul></li></ul></div>", "_____no_output_____" ], [ "# Feature engineering - quantifying access to facilities - batch mode\nThis notebook is similar to previous (04_feature-engineering-neighboring-facilities), except, this one runs for all shortlisted facilities and adds these as features.", "_____no_output_____" ] ], [ [ "import pandas as pd\nimport matplotlib.pyplot as plt\nfrom pprint import pprint\n%matplotlib inline\n\nfrom arcgis.gis import GIS\nfrom arcgis.geocoding import geocode, batch_geocode\nfrom arcgis.features import Feature, FeatureLayer, FeatureSet, GeoAccessor, GeoSeriesAccessor\nfrom arcgis.features import SpatialDataFrame\nfrom arcgis.geometry import Geometry, Point\nfrom arcgis.geometry.functions import buffer\nfrom arcgis.network import RouteLayer", "_____no_output_____" ] ], [ [ "Connect to GIS", "_____no_output_____" ] ], [ [ "gis = GIS(profile='')\nroute_service_url = gis.properties.helperServices.route.url\nroute_service = RouteLayer(route_service_url, gis=gis)", "_____no_output_____" ] ], [ [ "## Read the shortlisted properties", "_____no_output_____" ] ], [ [ "prop_list_df = pd.read_csv('resources/houses_for_sale_att_filtered.csv')\nprop_list_df.shape", "_____no_output_____" ], [ "prop_list_df = pd.DataFrame.spatial.from_xy(prop_list_df, 'LONGITUDE','LATITUDE')\ntype(prop_list_df)", "_____no_output_____" ] ], [ [ "## Loop through each property and build the neighborhood facility table", "_____no_output_____" ] ], [ [ "groceries_count = []\nrestaurants_count = []\nhospitals_count = []\ncoffee_count = []\nbars_count = []\ngas_count = []\nshops_service_count = []\ntravel_transport_count = []\nparks_count = []\neducation_count = []\nroute_length = []\nroute_duration = []\n\ndestination_address = '309 SW 6th Ave #600, Portland, OR 97204'", "_____no_output_____" ], [ "count=0\nfor index, prop in prop_list_df.iterrows():\n count+=1\n print(str(count), end=\": \")\n # geocode the property\n paddress = prop['ADDRESS'] + \", \" + prop['CITY'] + \", \" + prop['STATE']\n prop_geom_fset = geocode(paddress, as_featureset=True)\n \n print(prop['MLS'], end=\" : \")\n \n # create an envelope around each property\n prop_geom = prop_geom_fset.features[0]\n \n # create buffer of 5 miles\n prop_buffer = buffer([prop_geom.geometry], \n in_sr = 102100, buffer_sr=102100,\n distances=0.05, unit=9001)[0]\n\n prop_buffer_f = Feature(geometry=prop_buffer)\n prop_buffer_fset = FeatureSet([prop_buffer_f])\n \n # geocode for Groceries\n groceries = geocode('groceries', search_extent=prop_buffer.extent, \n max_locations=20, as_featureset=True)\n groceries_count.append(len(groceries.features))\n print('Groc', end=\" : \")\n \n # restaurants\n restaurants = geocode('restaurant', search_extent=prop_buffer.extent, max_locations=200)\n restaurants_count.append(len(restaurants))\n print('Rest', end=\" : \")\n \n # hospitals\n hospitals = geocode('hospital', search_extent=prop_buffer.extent, max_locations=50)\n hospitals_count.append(len(hospitals))\n print('Hosp', end =\" : \")\n \n # coffee shop\n coffees = geocode('coffee', search_extent=prop_buffer.extent, max_locations=50)\n coffee_count.append(len(coffees))\n print('Coffee', end=\" : \")\n \n # bars\n bars = geocode('bar', search_extent=prop_buffer.extent, max_locations=50)\n bars_count.append(len(bars))\n print('Bars', end=\" : \")\n \n # gas stations\n gas = geocode('gas station', search_extent=prop_buffer.extent, max_locations=50)\n gas_count.append(len(gas))\n print('Gas', end=\" : \")\n \n # shops\n shops_service = geocode(\"\",category='shops and service', \n search_extent=prop_buffer.extent, max_locations=50)\n shops_service_count.append(len(shops_service))\n print('Shops', end=\" : \")\n \n # travel & transport\n transport = geocode(\"\",category='travel and transport', \n search_extent=prop_buffer.extent, max_locations=50)\n travel_transport_count.append(len(transport))\n print(\"Travel\", end =\" : \")\n \n # parks\n parks = geocode(\"\",category='parks and outdoors', \n search_extent=prop_buffer.extent, max_locations=50)\n parks_count.append(len(parks))\n print('Parks', end=\" : \")\n \n # education\n education = geocode(\"\",category='education', search_extent=prop_buffer.extent, \n max_locations=50)\n education_count.append(len(education))\n print(\"Edu\", end=\" : \")\n \n # get route\n stops = [paddress, destination_address]\n stops_geocoded = batch_geocode(stops)\n\n stops_geocoded = [item['location'] for item in stops_geocoded]\n stops_geocoded2 = '{},{};{},{}'.format(stops_geocoded[0]['x'],stops_geocoded[0]['y'],\n stops_geocoded[1]['x'],stops_geocoded[1]['y'])\n\n route_result = route_service.solve(stops_geocoded2, return_routes=True, \n return_stops=False, return_directions=True,\n impedance_attribute_name='TravelTime',\n start_time=644511600000,\n return_barriers=False, return_polygon_barriers=False,\n return_polyline_barriers=False)\n route_length.append(route_result['directions'][0]['summary']['totalLength'])\n route_duration.append(route_result['directions'][0]['summary']['totalTime'])\n print(\"Route\")", "1: 18517652 : Groc : Rest : Hosp : Coffee : Bars : Gas : Shops : Travel : Parks : Edu : Route\n2: 18465613 : Groc : Rest : Hosp : Coffee : Bars : Gas : Shops : Travel : Parks : Edu : Route\n3: 18005102 : Groc : Rest : Hosp : Coffee : Bars : Gas : Shops : Travel : Parks : Edu : Route\n4: 18216924 : Groc : Rest : Hosp : Coffee : Bars : Gas : Shops : Travel : Parks : Edu : Route\n5: 18647164 : Groc : Rest : Hosp : Coffee : Bars : Gas : Shops : Travel : Parks : Edu : Route\n6: 18229660 : Groc : Rest : Hosp : Coffee : Bars : Gas : Shops : Travel : Parks : Edu : Route\n7: 18586790 : Groc : Rest : Hosp : Coffee : Bars : Gas : Shops : Travel : Parks : Edu : Route\n8: 18314898 : Groc : Rest : Hosp : Coffee : Bars : Gas : Shops : Travel : Parks : Edu : Route\n9: 18085278 : Groc : Rest : Hosp : Coffee : Bars : Gas : Shops : Travel : Parks : Edu : Route\n10: 18406186 : Groc : Rest : Hosp : Coffee : Bars : Gas : Shops : Travel : Parks : Edu : Route\n11: 18020972 : Groc : Rest : Hosp : Coffee : Bars : Gas : Shops : Travel : Parks : Edu : Route\n12: 18283940 : Groc : Rest : Hosp : Coffee : Bars : Gas : Shops : Travel : Parks : Edu : Route\n13: 18166839 : Groc : Rest : Hosp : Coffee : Bars : Gas : Shops : Travel : Parks : Edu : Route\n14: 18390189 : Groc : Rest : Hosp : Coffee : Bars : Gas : Shops : Travel : Parks : Edu : Route\n15: 18281614 : Groc : Rest : Hosp : Coffee : Bars : Gas : Shops : Travel : Parks : Edu : Route\n16: 18159838 : Groc : Rest : Hosp : Coffee : Bars : Gas : Shops : Travel : Parks : Edu : Route\n17: 18697929 : Groc : Rest : Hosp : Coffee : Bars : Gas : Shops : Travel : Parks : Edu : Route\n18: 18184111 : Groc : Rest : Hosp : Coffee : Bars : Gas : Shops : Travel : Parks : Edu : Route\n19: 18381383 : Groc : Rest : Hosp : Coffee : Bars : Gas : Shops : Travel : Parks : Edu : Route\n20: 18036264 : Groc : Rest : Hosp : Coffee : Bars : Gas : Shops : Travel : Parks : Edu : Route\n21: 18352241 : Groc : Rest : Hosp : Coffee : Bars : Gas : Shops : Travel : Parks : Edu : Route\n22: 18415529 : Groc : Rest : Hosp : Coffee : Bars : Gas : Shops : Travel : Parks : Edu : Route\n23: 18052500 : Groc : Rest : Hosp : Coffee : Bars : Gas : Shops : Travel : Parks : Edu : Route\n24: 18615156 : Groc : Rest : Hosp : Coffee : Bars : Gas : Shops : Travel : Parks : Edu : Route\n25: 18663618 : Groc : Rest : Hosp : Coffee : Bars : Gas : Shops : Travel : Parks : Edu : Route\n26: 18524346 : Groc : Rest : Hosp : Coffee : Bars : Gas : Shops : Travel : Parks : Edu : Route\n27: 18287995 : Groc : Rest : Hosp : Coffee : Bars : Gas : Shops : Travel : Parks : Edu : Route\n28: 18496603 : Groc : Rest : Hosp : Coffee : Bars : Gas : Shops : Travel : Parks : Edu : Route\n29: 18306005 : Groc : Rest : Hosp : Coffee : Bars : Gas : Shops : Travel : Parks : Edu : Route\n30: 18017989 : Groc : Rest : Hosp : Coffee : Bars : Gas : Shops : Travel : Parks : Edu : Route\n31: 18630734 : Groc : Rest : Hosp : Coffee : Bars : Gas : Shops : Travel : Parks : Edu : Route\n32: 18362577 : Groc : Rest : Hosp : Coffee : Bars : Gas : Shops : Travel : Parks : Edu : Route\n33: 18185462 : Groc : Rest : Hosp : Coffee : Bars : Gas : Shops : Travel : Parks : Edu : Route\n34: 18525026 : Groc : Rest : Hosp : Coffee : Bars : Gas : Shops : Travel : Parks : Edu : Route\n35: 18268485 : Groc : Rest : Hosp : Coffee : Bars : Gas : Shops : Travel : Parks : Edu : Route\n36: 18077776 : Groc : Rest : Hosp : Coffee : Bars : Gas : Shops : Travel : Parks : Edu : Route\n37: 18404645 : Groc : Rest : Hosp : Coffee : Bars : Gas : Shops : Travel : Parks : Edu : Route\n38: 18543295 : Groc : Rest : Hosp : Coffee : Bars : Gas : Shops : Travel : Parks : Edu : Route\n39: 18268007 : Groc : Rest : Hosp : Coffee : Bars : Gas : Shops : Travel : Parks : Edu : Route\n40: 18565385 : Groc : Rest : Hosp : Coffee : Bars : Gas : Shops : Travel : Parks : Edu : Route\n41: 18327093 : Groc : Rest : Hosp : Coffee : Bars : Gas : Shops : Travel : Parks : Edu : Route\n42: 18300182 : Groc : Rest : Hosp : Coffee : Bars : Gas : Shops : Travel : Parks : Edu : Route\n43: 18130222 : Groc : Rest : Hosp : Coffee : Bars : Gas : Shops : Travel : Parks : Edu : Route\n44: 18390288 : Groc : Rest : Hosp : Coffee : Bars : Gas : Shops : Travel : Parks : Edu : Route\n45: 18244519 : Groc : Rest : Hosp : Coffee : Bars : Gas : Shops : Travel : Parks : Edu : Route\n46: 18533145 : Groc : Rest : Hosp : Coffee : Bars : Gas : Shops : Travel : Parks : Edu : Route\n47: 18317832 : Groc : Rest : Hosp : Coffee : Bars : Gas : Shops : Travel : Parks : Edu : Route\n48: 18625148 : Groc : Rest : Hosp : Coffee : Bars : Gas : Shops : Travel : Parks : Edu : Route\n49: 18046552 : Groc : Rest : Hosp : Coffee : Bars : Gas : Shops : Travel : Parks : Edu : Route\n50: 18613304 : Groc : Rest : Hosp : Coffee : Bars : Gas : Shops : Travel : Parks : Edu : Route\n51: 18035240 : Groc : Rest : Hosp : Coffee : Bars : Gas : Shops : Travel : Parks : Edu : Route\n52: 18170798 : Groc : Rest : Hosp : Coffee : Bars : Gas : Shops : Travel : Parks : Edu : Route\n53: 18479918 : Groc : Rest : Hosp : Coffee : Bars : Gas : Shops : Travel : Parks : Edu : Route\n54: 18134679 : Groc : Rest : Hosp : Coffee : Bars : Gas : Shops : Travel : Parks : Edu : Route\n55: 18175979 : Groc : Rest : Hosp : Coffee : Bars : Gas : Shops : Travel : Parks : Edu : Route\n56: 18489535 : Groc : Rest : Hosp : Coffee : Bars : Gas : Shops : Travel : Parks : Edu : Route\n57: 18136667 : Groc : Rest : Hosp : Coffee : Bars : Gas : Shops : Travel : Parks : Edu : Route\n58: 18330192 : Groc : Rest : Hosp : Coffee : Bars : Gas : Shops : Travel : Parks : Edu : Route\n59: 18035177 : Groc : Rest : Hosp : Coffee : Bars : Gas : Shops : Travel : Parks : Edu : Route\n60: 18117942 : Groc : Rest : Hosp : Coffee : Bars : Gas : Shops : Travel : Parks : Edu : Route\n61: 18467170 : Groc : Rest : Hosp : Coffee : Bars : Gas : Shops : Travel : Parks : Edu : Route\n62: 18021854 : Groc : Rest : Hosp : Coffee : Bars : Gas : Shops : Travel : Parks : Edu : Route\n63: 18453150 : Groc : Rest : Hosp : Coffee : Bars : Gas : Shops : Travel : Parks : Edu : Route\n64: 17228335 : Groc : Rest : Hosp : Coffee : Bars : Gas : Shops : Travel : Parks : Edu : Route\n65: 18385412 : Groc : Rest : Hosp : Coffee : Bars : Gas : Shops : Travel : Parks : Edu : Route\n66: 18303159 : Groc : Rest : Hosp : Coffee : Bars : Gas : Shops : Travel : Parks : Edu : Route\n67: 18260962 : Groc : Rest : Hosp : Coffee : Bars : Gas : Shops : Travel : Parks : Edu : Route\n68: 18077788 : Groc : Rest : Hosp : Coffee : Bars : Gas : Shops : Travel : Parks : Edu : Route\n69: 1443106 : Groc : Rest : Hosp : Coffee : Bars : Gas : Shops : Travel : Parks : Edu : Route\n70: 18363380 : Groc : Rest : Hosp : Coffee : Bars : Gas : Shops : Travel : Parks : Edu : Route\n71: 18038013 : Groc : Rest : Hosp : Coffee : Bars : Gas : Shops : Travel : Parks : Edu : Route\n72: 18592639 : Groc : Rest : Hosp : Coffee : Bars : Gas : Shops : Travel : Parks : Edu : Route\n73: 18031108 : Groc : Rest : Hosp : Coffee : Bars : Gas : Shops : Travel : Parks : Edu : Route\n74: 18529842 : Groc : Rest : Hosp : Coffee : Bars : Gas : Shops : Travel : Parks : Edu : Route\n75: 18346063 : Groc : Rest : Hosp : Coffee : Bars : Gas : Shops : Travel : Parks : Edu : Route\n76: 18058515 : Groc : Rest : Hosp : Coffee : Bars : Gas : Shops : Travel : Parks : Edu : Route\n77: 18312442 : Groc : Rest : Hosp : Coffee : Bars : Gas : Shops : Travel : Parks : Edu : Route\n78: 18062133 : Groc : Rest : Hosp : Coffee : Bars : Gas : Shops : Travel : Parks : Edu : Route\n79: 18292112 : Groc : Rest : Hosp : Coffee : Bars : Gas : Shops : Travel : Parks : Edu : Route\n80: 18021426 : Groc : Rest : Hosp : Coffee : Bars : Gas : Shops : Travel : Parks : Edu : Route\n81: 18243995 : Groc : Rest : Hosp : Coffee : Bars : Gas : Shops : Travel : Parks : Edu : Route\n82: 18225445 : Groc : Rest : Hosp : Coffee : Bars : Gas : Shops : Travel : Parks : Edu : Route\n83: 18294835 : Groc : Rest : Hosp : Coffee : Bars : Gas : Shops : Travel : Parks : Edu : Route\n84: 18036942 : Groc : Rest : Hosp : Coffee : Bars : Gas : Shops : Travel : Parks : Edu : Route\n85: 18164206 : Groc : Rest : Hosp : Coffee : Bars : Gas : Shops : Travel : Parks : Edu : Route\n86: 18112429 : Groc : Rest : Hosp : Coffee : Bars : Gas : Shops : Travel : Parks : Edu : Route\n87: 18401204 : Groc : Rest : Hosp : Coffee : Bars : Gas : Shops : Travel : Parks : Edu : Route\n88: 18185257 : Groc : Rest : Hosp : Coffee : Bars : Gas : Shops : Travel : Parks : Edu : Route\n89: 18056844 : Groc : Rest : Hosp : Coffee : Bars : Gas : Shops : Travel : Parks : Edu : Route\n90: 18158997 : Groc : Rest : Hosp : Coffee : Bars : Gas : Shops : Travel : Parks : Edu : Route\n91: 18089346 : Groc : Rest : Hosp : Coffee : Bars : Gas : Shops : Travel : Parks : Edu : Route\n92: 18432056 : Groc : Rest : Hosp : Coffee : Bars : Gas : Shops : Travel : Parks : Edu : Route\n93: 18248238 : Groc : Rest : Hosp : Coffee : Bars : Gas : Shops : Travel : Parks : Edu : Route\n94: 18042235 : Groc : Rest : Hosp : Coffee : Bars : Gas : Shops : Travel : Parks : Edu : Route\n95: 18376736 : Groc : Rest : Hosp : Coffee : Bars : Gas : Shops : Travel : Parks : Edu : Route\n96: 17468803 : Groc : Rest : Hosp : Coffee : Bars : Gas : Shops : Travel : Parks : Edu : Route\n97: 18002685 : Groc : Rest : Hosp : Coffee : Bars : Gas : Shops : Travel : Parks : Edu : Route\n98: 18609084 : Groc : Rest : Hosp : Coffee : Bars : Gas : Shops : Travel : Parks : Edu : Route\n99: 18115705 : Groc : Rest : Hosp : Coffee : Bars : Gas : Shops : Travel : Parks : Edu : Route\n100: 18197366 : Groc : Rest : Hosp : Coffee : Bars : Gas : Shops : Travel : Parks : Edu : Route\n101: 18367807 : Groc : Rest : Hosp : Coffee : Bars : Gas : Shops : Travel : Parks : Edu : Route\n102: 18289907 : Groc : Rest : Hosp : Coffee : Bars : Gas : Shops : Travel : Parks : Edu : Route\n103: 1456193 : Groc : Rest : Hosp : Coffee : Bars : Gas : Shops : Travel : Parks : Edu : Route\n104: 18047073 : Groc : Rest : Hosp : Coffee : Bars : Gas : Shops : Travel : Parks : Edu : Route\n105: 18379344 : Groc : Rest : Hosp : Coffee : Bars : Gas : Shops : Travel : Parks : Edu : Route\n106: 18487228 : Groc : Rest : Hosp : Coffee : Bars : Gas : Shops : Travel : Parks : Edu : Route\n107: 18123653 : Groc : Rest : Hosp : Coffee : Bars : Gas : Shops : Travel : Parks : Edu : Route\n108: 18310961 : Groc : Rest : Hosp : Coffee : Bars : Gas : Shops : Travel : Parks : Edu : Route\n109: 18671543 : Groc : Rest : Hosp : Coffee : Bars : Gas : Shops : Travel : Parks : Edu : Route\n110: 18384430 : Groc : Rest : Hosp : Coffee : Bars : Gas : Shops : Travel : Parks : Edu : Route\n111: 18292179 : Groc : Rest : Hosp : Coffee : Bars : Gas : Shops : Travel : Parks : Edu : Route\n112: 18545728 : Groc : Rest : Hosp : Coffee : Bars : Gas : Shops : Travel : Parks : Edu : Route\n113: 18317933 : Groc : Rest : Hosp : Coffee : Bars : Gas : Shops : Travel : Parks : Edu : Route\n114: 18131614 : Groc : Rest : Hosp : Coffee : Bars : Gas : Shops : Travel : Parks : Edu : Route\n115: 18359207 : Groc : Rest : Hosp : Coffee : Bars : Gas : Shops : Travel : Parks : Edu : Route\n116: 18188909 : Groc : Rest : Hosp : Coffee : Bars : Gas : Shops : Travel : Parks : Edu : Route\n117: 18257295 : Groc : Rest : Hosp : Coffee : Bars : Gas : Shops : Travel : Parks : Edu : Route\n118: 18454489 : Groc : Rest : Hosp : Coffee : Bars : Gas : Shops : Travel : Parks : Edu : Route\n119: 18644320 : Groc : Rest : Hosp : Coffee : Bars : Gas : Shops : Travel : Parks : Edu : Route\n120: 18103930 : Groc : Rest : Hosp : Coffee : Bars : Gas : Shops : Travel : Parks : Edu : Route\n121: 18213089 : Groc : Rest : Hosp : Coffee : Bars : Gas : Shops : Travel : Parks : Edu : Route\n122: 18436725 : Groc : Rest : Hosp : Coffee : Bars : Gas : Shops : Travel : Parks : Edu : Route\n123: 18217493 : Groc : Rest : Hosp : Coffee : Bars : Gas : Shops : Travel : Parks : Edu : Route\n124: 18534351 : Groc : Rest : Hosp : Coffee : Bars : Gas : Shops : Travel : Parks : Edu : Route\n125: 18391773 : Groc : Rest : Hosp : Coffee : Bars : Gas : Shops : Travel : Parks : Edu : Route\n126: 17138243 : Groc : Rest : Hosp : Coffee : Bars : Gas : Shops : Travel : Parks : Edu : Route\n127: 18580547 : Groc : Rest : Hosp : Coffee : Bars : Gas : Shops : Travel : Parks : Edu : Route\n128: 18385584 : Groc : Rest : Hosp : Coffee : Bars : Gas : Shops : Travel : Parks : Edu : Route\n129: 1516748 : Groc : Rest : Hosp : Coffee : Bars : Gas : Shops : Travel : Parks : Edu : Route\n130: 18108097 : Groc : Rest : Hosp : Coffee : Bars : Gas : Shops : Travel : Parks : Edu : Route\n131: 18343244 : Groc : Rest : Hosp : Coffee : Bars : Gas : Shops : Travel : Parks : Edu : Route\n132: 18518261 : Groc : Rest : Hosp : Coffee : Bars : Gas : Shops : Travel : Parks : Edu : Route\n133: 18496090 : Groc : Rest : Hosp : Coffee : Bars : Gas : Shops : Travel : Parks : Edu : Route\n134: 18084407 : Groc : Rest : Hosp : Coffee : Bars : Gas : Shops : Travel : Parks : Edu : Route\n135: 18622197 : Groc : Rest : Hosp : Coffee : Bars : Gas : Shops : Travel : Parks : Edu : Route\n136: 18590953 : Groc : Rest : Hosp : Coffee : Bars : Gas : Shops : Travel : Parks : Edu : Route\n137: 18548575 : Groc : Rest : Hosp : Coffee : Bars : Gas : Shops : Travel : Parks : Edu : Route\n138: 18361209 : Groc : Rest : Hosp : Coffee : Bars : Gas : Shops : Travel : Parks : Edu : Route\n139: 18190102 : Groc : Rest : Hosp : Coffee : Bars : Gas : Shops : Travel : Parks : Edu : Route\n140: 18306032 : Groc : Rest : Hosp : Coffee : Bars : Gas : Shops : Travel : Parks : Edu : Route\n141: 18546456 : Groc : Rest : Hosp : Coffee : Bars : Gas : Shops : Travel : Parks : Edu : Route\n142: 17218625 : Groc : Rest : Hosp : Coffee : Bars : Gas : Shops : Travel : Parks : Edu : Route\n143: 18360477 : Groc : Rest : Hosp : Coffee : Bars : Gas : Shops : Travel : Parks : Edu : Route\n144: 18447320 : Groc : Rest : Hosp : Coffee : Bars : Gas : Shops : Travel : Parks : Edu : Route\n145: 18127206 : Groc : Rest : Hosp : Coffee : Bars : Gas : Shops : Travel : Parks : Edu : Route\n146: 18489822 : Groc : Rest : Hosp : Coffee : Bars : Gas : Shops : Travel : Parks : Edu : Route\n147: 18218285 : Groc : Rest : Hosp : Coffee : Bars : Gas : Shops : Travel : Parks : Edu : Route\n148: 18647011 : Groc : Rest : Hosp : Coffee : Bars : Gas : Shops : Travel : Parks : Edu : Route\n149: 18193927 : Groc : Rest : Hosp : Coffee : Bars : Gas : Shops : Travel : Parks : Edu : Route\n150: 18595550 : Groc : Rest : Hosp : Coffee : Bars : Gas : Shops : Travel : Parks : Edu : Route\n151: 18317847 : Groc : Rest : Hosp : Coffee : Bars : Gas : Shops : Travel : Parks : Edu : Route\n152: 18108604 : Groc : Rest : Hosp : Coffee : Bars : Gas : Shops : Travel : Parks : Edu : Route\n153: 18410442 : Groc : Rest : Hosp : Coffee : Bars : Gas : Shops : Travel : Parks : Edu : Route\n154: 17515218 : Groc : Rest : Hosp : Coffee : Bars : Gas : Shops : Travel : Parks : Edu : Route\n155: 18578497 : Groc : Rest : Hosp : Coffee : Bars : Gas : Shops : Travel : Parks : Edu : Route\n156: 18293042 : Groc : Rest : Hosp : Coffee : Bars : Gas : Shops : Travel : Parks : Edu : Route\n157: 18401105 : Groc : Rest : Hosp : Coffee : Bars : Gas : Shops : Travel : Parks : Edu : Route\n158: 18095097 : Groc : Rest : Hosp : Coffee : Bars : Gas : Shops : Travel : Parks : Edu : Route\n159: 18030852 : Groc : Rest : Hosp : Coffee : Bars : Gas : Shops : Travel : Parks : Edu : Route\n160: 18299685 : Groc : Rest : Hosp : Coffee : Bars : Gas : Shops : Travel : Parks : Edu : Route\n161: 18583483 : Groc : Rest : Hosp : Coffee : Bars : Gas : Shops : Travel : Parks : Edu : Route\n162: 18171268 : Groc : Rest : Hosp : Coffee : Bars : Gas : Shops : Travel : Parks : Edu : Route\n163: 18053604 : Groc : Rest : Hosp : Coffee : Bars : Gas : Shops : Travel : Parks : Edu : Route\n164: 18622831 : Groc : Rest : Hosp : Coffee : Bars : Gas : Shops : Travel : Parks : Edu : Route\n165: 18586722 : Groc : Rest : Hosp : Coffee : Bars : Gas : Shops : Travel : Parks : Edu : Route\n166: 18467321 : Groc : Rest : Hosp : Coffee : Bars : Gas : Shops : Travel : Parks : Edu : Route\n167: 18383920 : Groc : Rest : Hosp : Coffee : Bars : Gas : Shops : Travel : Parks : Edu : Route\n168: 18599482 : Groc : Rest : Hosp : Coffee : Bars : Gas : Shops : Travel : Parks : Edu : Route\n169: 18500611 : Groc : Rest : Hosp : Coffee : Bars : Gas : Shops : Travel : Parks : Edu : Route\n170: 18345364 : Groc : Rest : Hosp : Coffee : Bars : Gas : Shops : Travel : Parks : Edu : Route\n171: 18116411 : Groc : Rest : Hosp : Coffee : Bars : Gas : Shops : Travel : Parks : Edu : Route\n" ] ], [ [ "## Feature engineer with access to amenities", "_____no_output_____" ] ], [ [ "prop_list_df['grocery_count'] = groceries_count\nprop_list_df['restaurant_count']= restaurants_count\nprop_list_df['hospitals_count']= hospitals_count\nprop_list_df['coffee_count']= coffee_count\nprop_list_df['bars_count']=bars_count\nprop_list_df['gas_count']=gas_count\nprop_list_df['shops_count']=shops_service_count\nprop_list_df['travel_count']=travel_transport_count\nprop_list_df['parks_count']=parks_count\nprop_list_df['edu_count']=education_count\nprop_list_df['commute_length']=route_length\nprop_list_df['commute_duration']=route_duration\nprop_list_df.head()", "_____no_output_____" ] ], [ [ "### Plot the distribution of facility access", "_____no_output_____" ] ], [ [ "prop_list_df.columns", "_____no_output_____" ], [ "facility_list = ['grocery_count', 'restaurant_count', 'hospitals_count', 'coffee_count',\n 'bars_count', 'gas_count', 'shops_count', 'travel_count', 'parks_count',\n 'edu_count', 'commute_length', 'commute_duration']\n\naxes = prop_list_df[facility_list].hist(bins=25, layout=(3,4), figsize=(15,10))", "_____no_output_____" ] ], [ [ "From the histograms above, most houses don't have very many bars in 5 miles around them. The commute length and duration appears to be tightly clustered around the lower end of the spectrum. Most houses have at least 1 hospital or medical center near them and a large number of parks, restaurants, educational institutions.", "_____no_output_____" ], [ "### Store to disk", "_____no_output_____" ] ], [ [ "prop_list_df.to_csv('resources/houses_facility_counts.csv')\nprop_list_df.spatial.to_featureclass('resources/shp/houses_facility_counts.shp')", "_____no_output_____" ] ] ]

[ "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code" ]

[ [ "markdown", "markdown" ], [ "code" ], [ "markdown" ], [ "code" ], [ "markdown" ], [ "code", "code" ], [ "markdown" ], [ "code", "code" ], [ "markdown" ], [ "code" ], [ "markdown" ], [ "code", "code" ], [ "markdown", "markdown" ], [ "code" ] ]

[ "Apache-2.0" ]

[ "Apache-2.0" ]

[ "Apache-2.0" ]

[ [ [ "<a href=\"https://colab.research.google.com/github/Ishitha2003/-fmml20211052/blob/main/module1_lab_2.ipynb\" target=\"_parent\"><img src=\"https://colab.research.google.com/assets/colab-badge.svg\" alt=\"Open In Colab\"/></a>", "_____no_output_____" ], [ "# **The aim of this lab is to introduce DATA and FEATURES.**", "_____no_output_____" ], [ "Extracting features from data \nFMML Module 1, Lab 2 \nModule Coordinator : [email protected]", "_____no_output_____" ] ], [ [ "! pip install wikipedia\n\nimport wikipedia\nimport nltk\nfrom nltk.util import ngrams\nfrom collections import Counter\nimport matplotlib.pyplot as plt\nimport numpy as np\nimport re\nimport unicodedata\nimport plotly.express as px\nimport pandas as pd", "Collecting wikipedia\n Downloading wikipedia-1.4.0.tar.gz (27 kB)\nRequirement already satisfied: beautifulsoup4 in /usr/local/lib/python3.7/dist-packages (from wikipedia) (4.6.3)\nRequirement already satisfied: requests<3.0.0,>=2.0.0 in /usr/local/lib/python3.7/dist-packages (from wikipedia) (2.23.0)\nRequirement already satisfied: urllib3!=1.25.0,!=1.25.1,<1.26,>=1.21.1 in /usr/local/lib/python3.7/dist-packages (from requests<3.0.0,>=2.0.0->wikipedia) (1.24.3)\nRequirement already satisfied: idna<3,>=2.5 in /usr/local/lib/python3.7/dist-packages (from requests<3.0.0,>=2.0.0->wikipedia) (2.10)\nRequirement already satisfied: chardet<4,>=3.0.2 in /usr/local/lib/python3.7/dist-packages (from requests<3.0.0,>=2.0.0->wikipedia) (3.0.4)\nRequirement already satisfied: certifi>=2017.4.17 in /usr/local/lib/python3.7/dist-packages (from requests<3.0.0,>=2.0.0->wikipedia) (2021.10.8)\nBuilding wheels for collected packages: wikipedia\n Building wheel for wikipedia (setup.py) ... \u001b[?25l\u001b[?25hdone\n Created wheel for wikipedia: filename=wikipedia-1.4.0-py3-none-any.whl size=11695 sha256=1bd6bc02a88d2783cdab8fa979ba87dd77c590685818e4995e42f62ac8980708\n Stored in directory: /root/.cache/pip/wheels/15/93/6d/5b2c68b8a64c7a7a04947b4ed6d89fb557dcc6bc27d1d7f3ba\nSuccessfully built wikipedia\nInstalling collected packages: wikipedia\nSuccessfully installed wikipedia-1.4.0\n" ] ], [ [ "# **What are features?**\n\nfeatures are individual independent variables that act like a input to your system.", "_____no_output_____" ] ], [ [ "import matplotlib.pyplot as plt\nfrom matplotlib import cm\nimport numpy as np\n\nfrom mpl_toolkits.mplot3d.axes3d import get_test_data\n\n \n# set up a figure twice as wide as it is tall\nfig = plt.figure(figsize=plt.figaspect(0.9))\n\n# =============\n# First subplot\n# =============\n# set up the axes for the first plot\nax = fig.add_subplot(1, 2, 1, projection='3d')\n\n# plot a 3D surface like in the example mplot3d/surface3d_demo\nX = np.arange(-5, 5, 0.25) # feature 1\nY = np.arange(-5, 5, 0.25) # feature 2\nX, Y = np.meshgrid(X, Y)\nR = np.sqrt(X**2 + Y**2)\nZ = np.sin(R) #output\nsurf = ax.plot_surface(X, Y, Z, rstride=1, cstride=1, cmap=cm.coolwarm,\n linewidth=0.4, antialiased=False)\nax.set_zlim(-1.01, 1.01)\nfig.colorbar(surf, shrink=0.5, aspect=10)", "_____no_output_____" ] ], [ [ "# **Part 2: Features of text**\n\nHow do we apply machine learning on text? We can't directly use the text as input to our algorithms. We need to convert them to features. In this notebook, we will explore a simple way of converting text to features.\n\nLet us download a few documents off Wikipedia.", "_____no_output_____" ] ], [ [ "topic1 = 'Giraffe'\ntopic2 = 'Elephant'\nwikipedia.set_lang('en') \neng1 = wikipedia.page(topic1).content\neng2 = wikipedia.page(topic2).content\nwikipedia.set_lang('fr')\nfr1 = wikipedia.page(topic1).content\nfr2 = wikipedia.page(topic2).content\n", "_____no_output_____" ], [ "fr2", "_____no_output_____" ] ], [ [ "We need to clean this up a bit. Let us remove all the special characters and keep only 26 letters and space. Note that this will remove accented characters in French also. We are also removing all the numbers and spaces. So this is not an ideal solution.", "_____no_output_____" ] ], [ [ "def cleanup(text):\n text = text.lower() # make it lowercase\n text = re.sub('[^a-z]+', '', text) # only keep characters\n return text", "_____no_output_____" ], [ "print(eng1)", "The giraffe is a tall African mammal belonging to the genus Giraffa. Specifically, It is an even-toed ungulate. It is the tallest living terrestrial animal and the largest ruminant on Earth. Traditionally, giraffes were thought to be one species, Giraffa camelopardalis, with nine subspecies. Most recently, researchers proposed dividing giraffes into up to eight extant species due to new research into their mitochondrial and nuclear DNA, as well as morphological measurements. Seven other extinct species of Giraffa are known from the fossil record.\nThe giraffe's chief distinguishing characteristics are its extremely long neck and legs, its horn-like ossicones, and its spotted coat patterns. It is classified under the family Giraffidae, along with its closest extant relative, the okapi. Its scattered range extends from Chad in the north to South Africa in the south, and from Niger in the west to Somalia in the east. Giraffes usually inhabit savannahs and woodlands. Their food source is leaves, fruits, and flowers of woody plants, primarily acacia species, which they browse at heights most other herbivores cannot reach.\nLions, leopards, spotted hyenas, and African wild dogs may prey upon giraffes. Giraffes live in herds of related females and their offspring, or bachelor herds of unrelated adult males, but are gregarious and may gather in large aggregations. Males establish social hierarchies through \"necking,” which are combat bouts where the neck is used as a weapon. Dominant males gain mating access to females, which bear the sole responsibility for raising the young.\nThe giraffe has intrigued various ancient and modern cultures for its peculiar appearance, and has often been featured in paintings, books, and cartoons. It is classified by the International Union for Conservation of Nature (IUCN) as vulnerable to extinction and has been extirpated from many parts of its former range. Giraffes are still found in numerous national parks and game reserves, but estimates as of 2016 indicate there are approximately 97,500 members of Giraffa in the wild. More than 1,600 were kept in zoos in 2010.\n\n\n== Etymology ==\nThe name \"giraffe\" has its earliest known origins in the Arabic word zarāfah (زرافة), perhaps borrowed from the animal's Somali name geri. The Arab name is translated as \"fast-walker\". In early Modern English the spellings jarraf and ziraph were used, probably directly from the Arabic, and in Middle English orafle and gyrfaunt, gerfaunt. The Italian form giraffa arose in the 1590s. The modern English form developed around 1600 from the French girafe.\"Camelopard\" is an archaic English name for the giraffe; it derives from the Ancient Greek καμηλοπάρδαλις (kamēlopárdalis), from κάμηλος (kámēlos), \"camel\", and πάρδαλις (párdalis), \"leopard\", referring to its camel-like shape and leopard-like colouration.\n\n\n== Taxonomy ==\nCarl Linnaeus originally classified living giraffes as one species in 1758. He gave it the binomial name Cervus camelopardalis. Morten Thrane Brünnich classified the genus Giraffa in 1762. The species name camelopardalis is from Latin.\n\n\n=== Evolution ===\nThe giraffe is one of only two living genera of the family Giraffidae in the order Artiodactyla, the other being the okapi. The family was once much more extensive, with over 10 fossil genera described. The elongation of the neck appears to have started early in the giraffe lineage. Comparisons between giraffes and their ancient relatives suggest vertebrae close to the skull lengthened earlier, followed by lengthening of vertebrae further down. One early giraffid ancestor was Canthumeryx which has been dated variously to have lived 25–20 million years ago (mya), 17–15 mya or 18–14.3 mya and whose deposits have been found in Libya. This animal was medium-sized, slender and antelope-like. Giraffokeryx appeared 15 mya on the Indian subcontinent and resembled an okapi or a small giraffe, and had a longer neck and similar ossicones. Giraffokeryx may have shared a clade with more massively built giraffids like Sivatherium and Bramatherium.Giraffids like Palaeotragus, Shansitherium and Samotherium appeared 14 mya and lived throughout Africa and Eurasia. These animals had bare ossicones and small cranial sinuses and were longer with broader skulls. Paleotragus resembled the okapi and may have been its ancestor. Others find that the okapi lineage diverged earlier, before Giraffokeryx. Samotherium was a particularly important transitional fossil in the giraffe lineage, as its cervical vertebrae were intermediate in length and structure between a modern giraffe and an okapi, and were more vertical than the okapi's. Bohlinia, which first appeared in southeastern Europe and lived 9–7 mya was likely a direct ancestor of the giraffe. Bohlinia closely resembled modern giraffes, having a long neck and legs and similar ossicones and dentition.\n\nBohlinia entered China and northern India in response to climate change. From there, the genus Giraffa evolved and, around 7 mya, entered Africa. Further climate changes caused the extinction of the Asian giraffes, while the African giraffes survived and radiated into several new species. Living giraffes appear to have arisen around 1 mya in eastern Africa during the Pleistocene. Some biologists suggest the modern giraffes descended from G. jumae; others find G. gracilis a more likely candidate. G. jumae was larger and more heavily built, while G. gracilis was smaller and more lightly built.The changes from extensive forests to more open habitats, which began 8 mya, are believed to be the main driver for the evolution of giraffes. During this time, tropical plants disappeared and were replaced by arid C4 plants, and a dry savannah emerged across eastern and northern Africa and western India. Some researchers have hypothesised that this new habitat coupled with a different diet, including acacia species, may have exposed giraffe ancestors to toxins that caused higher mutation rates and a higher rate of evolution. The coat patterns of modern giraffes may also have coincided with these habitat changes. Asian giraffes are hypothesised to have had more okapi-like colourations.The giraffe genome is around 2.9 billion base pairs in length compared to the 3.3 billion base pairs of the okapi. Of the proteins in giraffe and okapi genes, 19.4% are identical. The divergence of giraffe and okapi lineages dates to around 11.5 mya. A small group of regulatory genes in the giraffe appear to be responsible for the animal's stature and associated circulatory adaptations.\n\n\n=== Species and subspecies ===\n\nThe International Union for Conservation of Nature (IUCN) currently recognises only one species of giraffe with nine subspecies. During the 1900s, various taxonomies with two or three species were proposed. A 2007 study on the genetics of giraffes using mitochondrial DNA suggested at least six lineages could be recognised as species. A 2011 study using detailed analyses of the morphology of giraffes, and application of the phylogenetic species concept, described eight species of living giraffes. A 2016 study also concluded that living giraffes consist of multiple species. The researchers suggested the existence of four species, which have not exchanged genetic information between each other for 1 to 2 million years.A 2020 study showed that depending on the method chosen, different taxonomic hypotheses recognizing from two to six species can be considered for the genus Giraffa. That study also found that multi-species coalescent methods can lead to taxonomic over-splitting, as those methods delimit geographic structures rather than species. The three-species hypothesis, which recognises G. camelopardalis, G. giraffa, and G. tippelskirchi, is highly supported by phylogenetic analyses and also corroborated by most population genetic and multi-species coalescent analyses. A 2021 whole genome sequencing study suggests the existence of four distinct species and seven subspecies.The cladogram below shows the phylogenetic relationship between the four proposed species and seven subspecies based on the genome analysis. Note the eight lineages correspond to eight of the traditional subspecies in the one species hypothesis. The Rothschild giraffe is subsumed into G. camelopardalis camelopardalis.\n\nThe following table compares the different hypotheses for giraffe species. The description column shows the traditional nine subspecies in the one species hypothesis.\nThe first extinct species to be described was Giraffa sivalensis Falconer and Cautley 1843, a reevaluation of a vertebra that was initially described as a fossil of the living giraffe. While taxonomic opinion may be lacking on some names, the extinct species that have been published include:\nGiraffa gracilis\nGiraffa jumae\nGiraffa priscilla\nGiraffa pomeli\nGiraffa punjabiensis\nGiraffa pygmaea\nGiraffa sivalensis\nGiraffa stillei\n\n\n== Appearance and anatomy ==\n\nFully grown giraffes stand 4.3–5.7 m (14.1–18.7 ft) tall, with males taller than females. The average weight is 1,192 kg (2,628 lb) for an adult male and 828 kg (1,825 lb) for an adult female. Despite its long neck and legs, the giraffe's body is relatively short.: 66  The skin of a giraffe is mostly gray, or tan, and can reach a thickness of 20 mm (0.79 in).: 87  The 80–100 centimetres (31–39 in) long tail ends in a long, dark tuft of hair and is used as a defense against insects.: 94 The coat has dark blotches or patches, which can be orange, chestnut, brown, or nearly black, separated by light hair, usually white or cream coloured. Male giraffes become darker as they age. The coat pattern has been claimed to serve as camouflage in the light and shade patterns of savannah woodlands. When standing among trees and bushes, they are hard to see at even a few metres distance. However, adult giraffes move about to gain the best view of an approaching predator, relying on their size and ability to defend themselves rather than on camouflage, which may be more important for calves. Each individual giraffe has a unique coat pattern. Giraffe calves inherit some coat pattern traits from their mothers, and variation in some spot traits is correlated with neonatal survival. The skin underneath the blotches may serve as windows for thermoregulation, being sites for complex blood vessel systems and large sweat glands.The fur may give the animal chemical defense, as its parasite repellents give it a characteristic scent. At least 11 main aromatic chemicals are in the fur, although indole and 3-methylindole are responsible for most of the smell. Because the males have a stronger odour than the females, the odour may also have sexual function.\n\n\n=== Head ===\n\nBoth sexes have prominent horn-like structures called ossicones, formed from ossified cartilage, covered in skin and fused to the skull at the parietal bones. Being vascularised, the ossicones may have a role in thermoregulation, and are used in combat between males. Appearance is a reliable guide to the sex or age of a giraffe: the ossicones of females and young are thin and display tufts of hair on top, whereas those of adult males end in knobs and tend to be bald on top. Also, a median lump, which is more prominent in males, emerges at the front of the skull. Males develop calcium deposits that form bumps on their skulls as they age. Multiple sinuses lighten a giraffe's skull.: 103  However, as males age, their skulls become heavier and more club-like, helping them become more dominant in combat. The occipital condyles of the skull allow the animal to tilt its head straight up and grab food on the branches above with the tongue.: 103, 110 Located on both sides of the head, the giraffe's eyes give it good eyesight and a wide field of vision from its great height.: 85, 102  The eye is larger than in other ungulates, with a greater retinal surface area. Giraffes possibly see in colour: 85  and their senses of hearing and smell are sharp. The ears are movable: 95  and the nostrils are slit-shaped, which may be an adaptation against blowing sand.The giraffe's prehensile tongue is about 45 cm (18 in) long. It is black, perhaps to protect against sunburn and is useful for grasping foliage, and delicately removing leaves from branches.: 109–110  The giraffe's upper lip is prehensile and useful when foraging, and is covered in hair to protect against thorns. Papillae cover the tongue and the inside of the mouth. The upper jaw has a hard palate and lacks front teeth. The molars and premolars have a low-crowned, broad surface with an almost square cross-section.: 106 \n\n\n=== Legs, locomotion and posture ===\n\nA giraffe's front and back legs are about the same length. The radius and ulna of the front legs are articulated by the carpus, which, while structurally equivalent to the human wrist, functions as a knee. It appears that a suspensory ligament allows the lanky legs to support the animal's great weight. The hooves of large male giraffes reach a diameter of 31 cm × 23 cm (12.2 in × 9.1 in).: 98  The rear of each hoof is low, and the fetlock is close to the ground, allowing the foot to provide additional support for the animal's weight. Giraffes lack dewclaws and interdigital glands. The giraffe's pelvis, though relatively short, has an ilium that is outspread at the upper ends.A giraffe has only two gaits: walking and galloping. Walking is done by moving the legs on one side of the body, then doing the same on the other side. When galloping, the hind legs move around the front legs before the latter move forward, and the tail will curl up. The animal relies on the forward and backward motions of its head and neck to maintain balance and the counter momentum while galloping.: 327–29  The giraffe can reach a sprint speed of up to 60 km/h (37 mph), and can sustain 50 km/h (31 mph) for several kilometres. Giraffes would probably not be competent swimmers as their long legs would be highly cumbersome in the water, although they could possibly float. When swimming, the thorax would be weighed down by the front legs, making it difficult for the animal to move its neck and legs in harmony or keep its head above the water's surface.A giraffe rests by lying with its body on top of its folded legs.: 329  To lie down, the animal kneels on its front legs and then lowers the rest of its body. To get back up, it first gets on its front knees and shifts hindquarters onto its back feet. It then moves from kneeling to standing on its front legs and pulls the rest of its body upwards, swinging its head for balance.: 67  If the giraffe wants to bend down to drink, it either spreads its front legs or bends its knees. Studies in captivity found the giraffe sleeps intermittently around 4.6 hours per day, mostly at night. It usually sleeps lying down; however, standing sleeps have been recorded, particularly in older individuals. Intermittent short \"deep sleep\" phases while lying are characterised by the giraffe bending its neck backwards and resting its head on the hip or thigh, a position believed to indicate paradoxical sleep.\n\n\n=== Neck ===\nThe giraffe has an extremely elongated neck, which can be up to 2.4 m (7.9 ft) in length. Along the neck is a mane made of short, erect hairs. The neck typically rests at an angle of 50–60 degrees, though juveniles have straighter necks and rest at 70 degrees.: 94  The long neck results from a disproportionate lengthening of the cervical vertebrae, not from the addition of more vertebrae. Each cervical vertebra is over 28 cm (11 in) long.: 71  They comprise 52–54 per cent of the length of the giraffe's vertebral column, compared with the 27–33 percent typical of similar large ungulates, including the giraffe's closest living relative, the okapi. This elongation largely takes place after birth, perhaps because giraffe mothers would have a difficult time giving birth to young with the same neck proportions as adults. The giraffe's head and neck are held up by large muscles and a strengthened nuchal ligament, which are anchored by long dorsal spines on the anterior thoracic vertebrae, giving the animal a hump.\n\nThe giraffe's neck vertebrae have ball and socket joints.: 71  The point of articulation between the cervical and thoracic vertebrae of giraffes is shifted to lie between the first and second thoracic vertebrae (T1 and T2), unlike most other ruminants where the articulation is between the seventh cervical vertebra (C7) and T1. This allows C7 to contribute directly to increased neck length and has given rise to the suggestion that T1 is actually C8, and that giraffes have added an extra cervical vertebra. However, this proposition is not generally accepted, as T1 has other morphological features, such as an articulating rib, deemed diagnostic of thoracic vertebrae, and because exceptions to the mammalian limit of seven cervical vertebrae are generally characterised by increased neurological anomalies and maladies.There are several hypotheses regarding the evolutionary origin and maintenance of elongation in giraffe necks. Charles Darwin originally suggested the \"competing browsers hypothesis\", which has been challenged only recently. It suggests that competitive pressure from smaller browsers, like kudu, steenbok and impala, encouraged the elongation of the neck, as it enabled giraffes to reach food that competitors could not. This advantage is real, as giraffes can and do feed up to 4.5 m (15 ft) high, while even quite large competitors, such as kudu, can feed up to only about 2 m (6 ft 7 in) high. There is also research suggesting that browsing competition is intense at lower levels, and giraffes feed more efficiently (gaining more leaf biomass with each mouthful) high in the canopy. However, scientists disagree about just how much time giraffes spend feeding at levels beyond the reach of other browsers,\nand a 2010 study found that adult giraffes with longer necks actually suffered higher mortality rates under drought conditions than their shorter-necked counterparts. This study suggests that maintaining a longer neck requires more nutrients, which puts longer-necked giraffes at risk during a food shortage.Another theory, the sexual selection hypothesis, proposes the long necks evolved as a secondary sexual characteristic, giving males an advantage in \"necking\" contests (see below) to establish dominance and obtain access to sexually receptive females. In support of this theory, necks are longer and heavier for males than females of the same age, and males do not employ other forms of combat. However, one objection is it fails to explain why female giraffes also have long necks. It has also been proposed that the neck serves to give the animal greater vigilance.\n\n\n=== Internal systems ===\n\nIn mammals, the left recurrent laryngeal nerve is longer than the right; in the giraffe, it is over 30 cm (12 in) longer. These nerves are longer in the giraffe than in any other living animal; the left nerve is over 2 m (6 ft 7 in) long. Each nerve cell in this path begins in the brainstem and passes down the neck along the vagus nerve, then branches off into the recurrent laryngeal nerve which passes back up the neck to the larynx. Thus, these nerve cells have a length of nearly 5 m (16 ft) in the largest giraffes. Despite its long neck and large skull, the brain of the giraffe is typical for an ungulate. Evaporative heat loss in the nasal passages keep the giraffe's brain cool. The shape of the skeleton gives the giraffe a small lung volume relative to its mass. Its long neck gives it a large amount of dead space, in spite of its narrow windpipe. The giraffe also has an increased level of tidal volume so the ratio of dead space to tidal volume is similar to other mammals. The animal can still supply enough oxygen to its tissues, and it can increase its respiratory rate and oxygen diffusion when running.\n\nThe circulatory system of the giraffe has several adaptations for its great height. Its heart, which can weigh more than 11 kg (25 lb) and measures about 60 cm (2 ft) long, must generate approximately double the blood pressure required for a human to maintain blood flow to the brain. As such, the wall of the heart can be as thick as 7.5 cm (3.0 in). Giraffes have unusually high heart rates for their size, at 150 beats per minute.: 76  When the animal lowers its head, the blood rushes down fairly unopposed and a rete mirabile in the upper neck, with its large cross-sectional area, prevents excess blood flow to the brain. When it raises again, the blood vessels constrict and direct blood into the brain so the animal does not faint. The jugular veins contain several (most commonly seven) valves to prevent blood flowing back into the head from the inferior vena cava and right atrium while the head is lowered. Conversely, the blood vessels in the lower legs are under great pressure because of the weight of fluid pressing down on them. To solve this problem, the skin of the lower legs is thick and tight, preventing too much blood from pouring into them.Giraffes have oesophageal muscles that are unusually strong to allow regurgitation of food from the stomach up the neck and into the mouth for rumination.: 78  They have four chambered stomachs, as in all ruminants; the first chamber has adapted to their specialized diet. The intestines of an adult giraffe measure more than 70 m (230 ft) in length and have a relatively small ratio of small to large intestine. The liver of the giraffe is small and compact.: 76  A gallbladder is generally present during fetal life, but it may disappear before birth.\n\n\n== Behaviour and ecology ==\n\n\n=== Habitat and feeding ===\n\nGiraffes usually inhabit savannahs and open woodlands. They prefer Acacieae, Commiphora, Combretum and open Terminalia woodlands over denser environments like Brachystegia woodlands.: 322  The Angolan giraffe can be found in desert environments. Giraffes browse on the twigs of trees, preferring those of the subfamily Acacieae and the genera Commiphora and Terminalia, which are important sources of calcium and protein to sustain the giraffe's growth rate. They also feed on shrubs, grass and fruit.: 324  A giraffe eats around 34 kg (75 lb) of foliage daily. When stressed, giraffes may chew the bark off branches.: 325  Giraffes are also recorded to chew old bones.: 102 During the wet season, food is abundant and giraffes are more spread out, while during the dry season, they gather around the remaining evergreen trees and bushes. Mothers tend to feed in open areas, presumably to make it easier to detect predators, although this may reduce their feeding efficiency. As a ruminant, the giraffe first chews its food, then swallows it for processing and then visibly passes the half-digested cud up the neck and back into the mouth to chew again.: 78–79  The giraffe requires less food than many other herbivores because the foliage it eats has more concentrated nutrients and it has a more efficient digestive system. The animal's faeces come in the form of small pellets. When it has access to water, a giraffe drinks at intervals no longer than three days.Giraffes have a great effect on the trees that they feed on, delaying the growth of young trees for some years and giving \"waistlines\" to too tall trees. Feeding is at its highest during the first and last hours of daytime. Between these hours, giraffes mostly stand and ruminate. Rumination is the dominant activity during the night, when it is mostly done lying down.\n\n\n=== Social life ===\n\nGiraffes are usually found in groups that vary in size and composition according to ecological, anthropogenic, temporal, and social factors. Traditionally, the composition of these groups had been described as open and ever-changing. For research purposes, a \"group\" has been defined as \"a collection of individuals that are less than a kilometre apart and moving in the same general direction\". More recent studies have found that giraffes have long-term social associations and may form groups or pairs based on kinship, sex or other factors, and these groups regularly associate with other groups in larger communities or sub-communities within a fission–fusion society. Proximity to humans can disrupt social arrangements. Masai giraffes of Tanzania live in distinct social subpopulations that overlap spatially, but have different reproductive rates and calf survival rates.\nThe number of giraffes in a group can range from one up to 66 individuals. Giraffe groups tend to be sex-segregated although mixed-sex groups made of adult females and young males also occur. Female groups may be matrilineally related. Generally females are more selective than males in who they associate with regarding individuals of the same sex. Particularly stable giraffe groups are those made of mothers and their young, which can last weeks or months. Young males also form groups and will engage in playfights. However, as they get older, males become more solitary but may also associate in pairs or with female groups. Giraffes are not territorial, but they have home ranges that vary according to rainfall and proximity to human settlements. Male giraffes occasionally wander far from areas that they normally frequent.: 329 Early biologists suggested giraffes were mute and unable to produce air flow of sufficient velocity to vibrate their vocal folds. To the contrary; they have been recorded to communicate using snorts, sneezes, coughs, snores, hisses, bursts, moans, grunts, growls and flute-like sounds. During courtship, males emit loud coughs. Females call their young by bellowing. Calves will emit snorts, bleats, mooing and mewing sounds. Snorting and hissing in adults is associated with vigilance. During nighttime, giraffes appear to hum to each other above the infrasound range. The purpose is unclear. Dominant males display to other males with an erect posture; holding the chin and head high while walking stiffly and approaching them laterally. The less dominant show submissiveness by lowing the head and ears with the chin moved in and then jump and flee.\n\n\n=== Reproduction and parental care ===\n\nReproduction in giraffes is broadly polygamous: a few older males mate with the fertile females. Females can reproduce throughout the year and experience oestrus cycling approximately every 15 days. Female giraffes in oestrous are dispersed over space and time, so reproductive adult males adopt a strategy of roaming among female groups to seek mating opportunities, with periodic hormone-induced rutting behaviour approximately every two weeks. Males prefer young adult females over juveniles and older adults.Male giraffes assess female fertility by tasting the female's urine to detect oestrus, in a multi-step process known as the flehmen response. Once an oestrous female is detected, the male will attempt to court her. When courting, dominant males will keep subordinate ones at bay. A courting male may lick a female's tail, rest his head and neck on her body or nudge her with his ossicones. During copulation, the male stands on his hind legs with his head held up and his front legs resting on the female's sides.Giraffe gestation lasts 400–460 days, after which a single calf is normally born, although twins occur on rare occasions. The mother gives birth standing up. The calf emerges head and front legs first, having broken through the fetal membranes, and falls to the ground, severing the umbilical cord. A newborn giraffe is 1.7–2 m (5.6–6.6 ft) tall. Within a few hours of birth, the calf can run around and is almost indistinguishable from a one-week-old. However, for the first one to three weeks, it spends most of its time hiding; its coat pattern providing camouflage. The ossicones, which have lain flat while it was in the womb, become erect within a few days.\n\nMothers with calves will gather in nursery herds, moving or browsing together. Mothers in such a group may sometimes leave their calves with one female while they forage and drink elsewhere. This is known as a \"calving pool\". Adult males play almost no role in raising the young,: 337  although they appear to have friendly interactions. Calves are at risk of predation, and a mother giraffe will stand over her calf and kick at an approaching predator. Females watching calving pools will only alert their own young if they detect a disturbance, although the others will take notice and follow. Calves may be weaned at six to eight months old but can remain with their mothers for up to 14 months.: 49  Females become sexually mature when they are four years old, while males become mature at four or five years. Spermatogenesis in male giraffes begins at three to four years of age. Males must wait until they are at least seven years old to gain the opportunity to mate.\n\n\n=== Necking ===\n\nMale giraffes use their necks as weapons in combat, a behaviour known as \"necking\". Necking is used to establish dominance and males that win necking bouts have greater reproductive success. This behaviour occurs at low or high intensity. In low-intensity necking, the combatants rub and lean against each other. The male that can hold itself more erect wins the bout. In high-intensity necking, the combatants will spread their front legs and swing their necks at each other, attempting to land blows with their ossicones. The contestants will try to dodge each other's blows and then get ready to counter. The power of a blow depends on the weight of the skull and the arc of the swing. A necking duel can last more than half an hour, depending on how well matched the combatants are.: 331  Although most fights do not lead to serious injury, there have been records of broken jaws, broken necks, and even deaths.After a duel, it is common for two male giraffes to caress and court each other. Such interactions between males have been found to be more frequent than heterosexual coupling. In one study, up to 94 percent of observed mounting incidents took place between males. The proportion of same-sex activities varied from 30 to 75 percent. Only one percent of same-sex mounting incidents occurred between females.\n\n\n=== Mortality and health ===\n\nGiraffes have high adult survival probability, and an unusually long lifespan compared to other ruminants, up to 38 years. Because of their size, eyesight and powerful kicks, adult giraffes are usually not subject to predation, although lions may regularly prey on individuals up to 550 kg (1,210 lb). Giraffes are the most common food source for the big cats in Kruger National Park, comprising nearly a third of the meat consumed, although only a small portion of the giraffes were probably killed by predators, as a majority of the consumed giraffes appeared to be scavenged. Adult female survival is significantly correlated with gregariousness, the average number of other females she is seen associating with. Calves are much more vulnerable than adults and are also preyed on by leopards, spotted hyenas and wild dogs. A quarter to a half of giraffe calves reach adulthood. Calf survival varies according to the season of birth, with calves born during the dry season having higher survival rates.The local, seasonal presence of large herds of migratory wildebeests and zebras reduces predation pressure on giraffe calves and increases their survival probability. In turn, it has been suggested that other ungulates may benefit from associating with giraffes, as their height allows them to spot predators from further away. Zebras were found to glean information on predation risk from giraffe body language and spend less time scanning the environment when giraffes are present.Some parasites feed on giraffes. They are often hosts for ticks, especially in the area around the genitals, which have thinner skin than other areas. Tick species that commonly feed on giraffes are those of genera Hyalomma, Amblyomma and Rhipicephalus. Giraffes may rely on red-billed and yellow-billed oxpeckers to clean them of ticks and alert them to danger. Giraffes host numerous species of internal parasites and are susceptible to various diseases. They were victims of the (now eradicated) viral illness rinderpest. Giraffes can also suffer from a skin disorder, which comes in the form of wrinkles, lesions or raw fissures. As much as 79% of giraffes show signs of the disease in Ruaha National Park, but it did not cause mortality in Tarangire and is less prevalent in areas with fertile soils.\n\n\n== Relationship with humans ==\n\n\n=== Cultural significance ===\nWith its lanky build and spotted coat, the giraffe has been a source of fascination throughout human history, and its image is widespread in culture. It has been used to symbolise flexibility, far-sightedness, femininity, fragility, passivity, grace, beauty and the continent of Africa itself.: 7, 116 \n\nGiraffes were depicted in art throughout the African continent, including that of the Kiffians, Egyptians, and Kushites.: 45–47  The Kiffians were responsible for a life-size rock engraving of two giraffes, dated 8,000 years ago, that has been called the \"world's largest rock art petroglyph\".: 45  How the giraffe got its height has been the subject of various African folktales. The Tugen people of modern Kenya used the giraffe to depict their god Mda. The Egyptians gave the giraffe its own hieroglyph, named 'sr' in Old Egyptian and 'mmy' in later periods.: 49 Giraffes have a presence in modern Western culture. Salvador Dalí depicted them with burning manes in some of his surrealist paintings. Dali considered the giraffe to be a symbol of masculinity, and a flaming giraffe was meant to be a \"masculine cosmic apocalyptic monster\".: 123  Several children's books feature the giraffe, including David A. Ufer's The Giraffe Who Was Afraid of Heights, Giles Andreae's Giraffes Can't Dance and Roald Dahl's The Giraffe and the Pelly and Me. Giraffes have appeared in animated films, as minor characters in Disney's The Lion King and Dumbo, and in more prominent roles in The Wild and the Madagascar films. Sophie the Giraffe has been a popular teether since 1961. Another famous fictional giraffe is the Toys \"R\" Us mascot Geoffrey the Giraffe.: 127 The giraffe has also been used for some scientific experiments and discoveries. Scientists have looked at the properties of giraffe skin when developing suits for astronauts and fighter pilots: 76  because the people in these professions are in danger of passing out if blood rushes to their legs. Computer scientists have modeled the coat patterns of several subspecies using reaction–diffusion mechanisms. The constellation of Camelopardalis, introduced in the seventeenth century, depicts a giraffe.: 119–20  The Tswana people of Botswana traditionally see the constellation Crux as two giraffes—Acrux and Mimosa forming a male, and Gacrux and Delta Crucis forming the female.\n\n\n=== Captivity ===\nThe Egyptians kept giraffes as pets and shipped them around the Mediterranean.: 48–49  The giraffe was among the many animals collected and displayed by the Romans. The first one in Rome was brought in by Julius Caesar in 46 BC and exhibited to the public.: 52  With the fall of the Western Roman Empire, the housing of giraffes in Europe declined.: 54  During the Middle Ages, giraffes were known to Europeans through contact with the Arabs, who revered the giraffe for its peculiar appearance.Individual captive giraffes were given celebrity status throughout history. In 1414, a giraffe was shipped from Malindi to Bengal. It was then taken to China by explorer Zheng He and placed in a Ming dynasty zoo. The animal was a source of fascination for the Chinese people, who associated it with the mythical Qilin.: 56  The Medici giraffe was a giraffe presented to Lorenzo de' Medici in 1486. It caused a great stir on its arrival in Florence. Zarafa, another famous giraffe, was brought from Egypt to Paris in the early 19th century as a gift from Muhammad Ali of Egypt to Charles X of France. A sensation, the giraffe was the subject of numerous memorabilia or \"giraffanalia\".: 81 Giraffes have become popular attractions in modern zoos, though keeping them healthy is difficult as they require wide areas and high amounts of browse for food. Captive giraffes in North America and Europe appear to have a higher mortality rate than in the wild; causes of death include poor husbandry, nutrition and management decisions.: 153  Giraffes in zoos display stereotypical behaviours, the most common being the licking of non-food items.: 164  Zookeepers may offer various activities to stimulate giraffes, including training them to accept food from visitors.: 175  Stables for giraffes are built particularly high to accommodate their height.: 183 \n\n\n=== Exploitation ===\nGiraffes were probably common targets for hunters throughout Africa. Different parts of their bodies were used for different purposes. Their meat was used for food. The tail hairs served as flyswatters, bracelets, necklaces, and thread. Shields, sandals, and drums were made using the skin, and the strings of musical instruments were from the tendons. The smoke from burning giraffe skins was used by the medicine men of Buganda to treat nose bleeds. The Humr people of Kordofan consume the drink Umm Nyolokh, which is prepared from the liver and bone marrow of giraffes. Richard Rudgley hypothesised that Umm Nyolokh might contain DMT. The drink is said to cause hallucinations of giraffes, believed to be the giraffes' ghosts, by the Humr.\n\n\n=== Conservation status ===\nIn 2016, giraffes were assessed as Vulnerable from a conservation perspective by the IUCN. In 1985, it was estimated there were 155,000 giraffes in the wild. This declined to over 140,000 in 1999. Estimates as of 2016 indicate there are approximately 97,500 members of Giraffa in the wild. The Masai and reticulated subspecies are endangered, and the Rothschild subspecies is near threatened. The Nubian subspecies is critically endangered.\n\nThe primary causes for giraffe population declines are habitat loss and direct killing for bushmeat markets. Giraffes have been extirpated from much of their historic range, including Eritrea, Guinea, Mauritania and Senegal. They may also have disappeared from Angola, Mali, and Nigeria, but have been introduced to Rwanda and Eswatini. As of 2010, there were more than 1,600 in captivity at Species360-registered zoos. Habitat destruction has hurt the giraffe. In the Sahel, the need for firewood and grazing room for livestock has led to deforestation. Normally, giraffes can coexist with livestock, since they do not directly compete with them. In 2017, severe droughts in northern Kenya led to increased tensions over land and the killing of wildlife by herders, with giraffe populations being particularly hit.Protected areas like national parks provide important habitat and anti-poaching protection to giraffe populations. Community-based conservation efforts outside national parks are also effective at protecting giraffes and their habitats. Private game reserves have contributed to the preservation of giraffe populations in southern Africa. The giraffe is a protected species in most of its range. It is the national animal of Tanzania, and is protected by law, and unauthorised killing can result in imprisonment. The UN backed Convention of Migratory Species selected giraffes for protection in 2017. In 2019, giraffes were listed under Appendix II of the Convention on International Trade in Endangered Species (CITES), which means international trade including in parts/derivatives is regulated.Translocations are sometimes used to augment or re-establish diminished or extirpated populations, but these activities are risky and difficult to undertake using the best practices of extensive pre- and post-translocation studies and ensuring a viable founding population. Aerial survey is the most common method of monitoring giraffe population trends in the vast roadless tracts of African landscapes, but aerial methods are known to undercount giraffes. Ground-based survey methods are more accurate and can be used in conjunction with aerial surveys to make accurate estimates of population sizes and trends. The Giraffe Conservation Foundation has been criticized for alleged mistreatment of giraffes and giraffe scientists.\n\n\n== See also ==\nFauna of Africa\nGiraffe Centre\nGiraffe Manor - hotel in Nairobi with giraffes\n\n\n== References ==\n\n\n== External links ==\nGiraffe Conservation Foundation\n" ] ], [ [ "Instead of directly using characters as the features, to understand a text better, we may consider group of tokens i.e. ngrams as features.\nfor this example let us consider that each character is one word, and let us see how n-grams work.\n", "_____no_output_____" ], [ "# **nltk library provides many tools for text processing, please explore them.**", "_____no_output_____" ], [ "Now let us calculate the frequency of the character n-grams. N-grams are groups of characters of size n. A unigram is a single character and a bigram is a group of two characters and so on.\n\nLet us count the frequency of each character in a text and plot it in a histogram.", "_____no_output_____" ] ], [ [ "# convert a tuple of characters to a string\ndef tuple2string(tup):\n st = ''\n for ii in tup:\n st = st + ii\n return st\n\n# convert a tuple of tuples to a list of strings\ndef key2string(keys):\n return [tuple2string(i) for i in keys]\n\n# plot the histogram\ndef plothistogram(ngram):\n keys = key2string(ngram.keys()) \n values = list(ngram.values())\n \n # sort the keys in alphabetic order\n combined = zip(keys, values)\n zipped_sorted = sorted(combined, key=lambda x: x[0])\n keys, values = map(list, zip(*zipped_sorted))\n plt.bar(keys, values)", "_____no_output_____" ] ], [ [ "Let us compare the histograms of English pages and French pages. Can you spot a difference?", "_____no_output_____" ] ], [ [ "## we passed ngrams 'n' as 1 to get unigrams. Unigram is nothing but single token (in this case character).\n\nunigram_eng1 = Counter(ngrams(eng1,1))\nplothistogram(unigram_eng1)\nplt.title('English 1')\nplt.show()\nunigram_eng2 = Counter(ngrams(eng2,1))\nplothistogram(unigram_eng2)\nplt.title('English 2')\nplt.show()", "_____no_output_____" ], [ "unigram_fr1 = Counter(ngrams(fr1,1))\nplothistogram(unigram_eng1)\nplt.title('French 1')\nplt.show()\nunigram_fr2 = Counter(ngrams(fr2,1))\nplothistogram(unigram_fr2)\nplt.title('French 2')\nplt.show()", "_____no_output_____" ] ], [ [ "A good feature is one that helps in easy prediction and classification.\nfor ex : if you wish to differentiate between grapes and apples, size can be one of the useful features.", "_____no_output_____" ], [ "We can see that the unigrams for French and English are very similar. So this is not a good feature if we want to distinguish between English and French. Let us look at bigrams.", "_____no_output_____" ] ], [ [ "## Now instead of unigram, we will use bigrams as features, and see how useful bigrams are as features.\n\nbigram_eng1 = Counter(ngrams(eng1,2)) # bigrams\nplothistogram(bigram_eng1)\nplt.title('English 1')\nplt.show()\n\nbigram_eng2 = Counter(ngrams(eng2,2))\nplothistogram(bigram_eng2)\nplt.title('English 2')\nplt.show()\n\nbigram_fr1 = Counter(ngrams(fr1,2))\nplothistogram(bigram_eng1)\nplt.title('French 1')\nplt.show()\n\nbigram_fr2 = Counter(ngrams(fr2,2))\nplothistogram(bigram_fr2)\nplt.title('French 2')\nplt.show()", "_____no_output_____" ] ], [ [ "Another way to visualize bigrams is to use a 2-dimensional graph.", "_____no_output_____" ] ], [ [ "## lets have a lot at bigrams.\n\nbigram_eng1", "_____no_output_____" ], [ "def plotbihistogram(ngram):\n freq = np.zeros((26,26))\n for ii in range(26):\n for jj in range(26):\n freq[ii,jj] = ngram[(chr(ord('a')+ii), chr(ord('a')+jj))]\n plt.imshow(freq, cmap = 'jet')\n return freq", "_____no_output_____" ], [ "bieng1 = plotbihistogram(bigram_eng1)\nplt.show()\nbieng2 = plotbihistogram(bigram_eng2)", "_____no_output_____" ], [ "bifr1 = plotbihistogram(bigram_fr1)\nplt.show()\nbifr2 = plotbihistogram(bigram_fr2)", "_____no_output_____" ] ], [ [ "Let us look at the top 10 ngrams for each text.", "_____no_output_____" ] ], [ [ "from IPython.core.debugger import set_trace\n\ndef ind2tup(ind):\n ind = int(ind)\n i = int(ind/26)\n j = int(ind%26)\n return (chr(ord('a')+i), chr(ord('a')+j))\n\ndef ShowTopN(bifreq, n=10):\n f = bifreq.flatten()\n arg = np.argsort(-f)\n for ii in range(n):\n print(f'{ind2tup(arg[ii])} : {f[arg[ii]]}')\n", "_____no_output_____" ], [ "print('\\nEnglish 1:')\nShowTopN(bieng1)\nprint('\\nEnglish 2:')\nShowTopN(bieng2)\nprint('\\nFrench 1:')\nShowTopN(bifr1)\nprint('\\nFrench 2:')\nShowTopN(bifr2)", "\nEnglish 1:\n('t', 'h') : 714.0\n('h', 'e') : 705.0\n('i', 'n') : 577.0\n('e', 's') : 546.0\n('a', 'n') : 541.0\n('e', 'r') : 457.0\n('r', 'e') : 445.0\n('r', 'a') : 418.0\n('a', 'l') : 407.0\n('n', 'd') : 379.0\n\nEnglish 2:\n('a', 'n') : 1344.0\n('t', 'h') : 1271.0\n('h', 'e') : 1163.0\n('i', 'n') : 946.0\n('e', 'r') : 744.0\n('l', 'e') : 707.0\n('r', 'e') : 704.0\n('n', 'd') : 670.0\n('n', 't') : 642.0\n('h', 'a') : 632.0\n\nFrench 1:\n('e', 's') : 546.0\n('l', 'e') : 337.0\n('d', 'e') : 328.0\n('e', 'n') : 315.0\n('o', 'n') : 306.0\n('n', 't') : 275.0\n('r', 'e') : 268.0\n('r', 'a') : 216.0\n('a', 'n') : 202.0\n('o', 'u') : 193.0\n\nFrench 2:\n('e', 's') : 920.0\n('n', 't') : 773.0\n('d', 'e') : 623.0\n('e', 'n') : 598.0\n('a', 'n') : 538.0\n('l', 'e') : 511.0\n('o', 'n') : 479.0\n('r', 'e') : 455.0\n('u', 'r') : 295.0\n('t', 'i') : 280.0\n" ] ], [ [ "## **At times, we need to reduce the number of features. We will discuss this more in the upcoming sessions, but a small example has been discussed here. Instead of using each unique token (a word) as a feature, we reduced the number of features by using 1-gram and 2-gram of characters as features.**", "_____no_output_____" ], [ "We observe that the bigrams are similar across different topics but different across languages. Thus, the bigram frequency is a good feature for distinguishing languages, but not for distinguishing topics.\n\nThus, we were able to convert a many-dimensional input (the text) to 26 dimesions (unigrams) or 26*26 dimensions (bigrams).\n\nA few ways to explore:\n\nTry with different languages.\nThe topics we used are quite similar, wikipedia articles of 'elephant' and 'giraffe'. What happens if we use very different topics? What if we use text from another source than Wikipedia?\nHow can we use and visualize trigrams and higher n-grams?", "_____no_output_____" ], [ "Features of Images.\nImages in digital format are stored as numeric values, and hence we can use these values as features. for ex : a black and white (binary) image is stored as an array of 0 and 255 or 0 and 1.", "_____no_output_____" ], [ "# **Part 2: Written numbers**\n\nWe will use a subset of the MNIST dataset. Each input character is represented in a 28*28 array. Let us see if we can extract some simple features from these images which can help us distinguish between the digits.\n\nLoad the dataset:", "_____no_output_____" ] ], [ [ "from keras.datasets import mnist\n \n#loading the dataset\n(train_X, train_y), (test_X, test_y) = mnist.load_data()", "Downloading data from https://storage.googleapis.com/tensorflow/tf-keras-datasets/mnist.npz\n11493376/11490434 [==============================] - 0s 0us/step\n11501568/11490434 [==============================] - 0s 0us/step\n" ] ], [ [ "Extract a subset of the data for our experiment:", "_____no_output_____" ] ], [ [ "no1 = train_X[train_y==1,:,:] ## dataset corresponding to number = 1.\nno0 = train_X[train_y==0,:,:] ## dataset corresponding to number = 0.", "_____no_output_____" ] ], [ [ "Let us visualize a few images here", "_____no_output_____" ] ], [ [ "for ii in range(5):\n plt.subplot(1, 5, ii+1)\n plt.imshow(no1[ii,:,:])\nplt.show()\nfor ii in range(5):\n plt.subplot(1, 5, ii+1)\n plt.imshow(no0[ii,:,:])\nplt.show()", "_____no_output_____" ] ], [ [ "We can even use value of each pixel as a feature. But let us see how to derive other features.\n", "_____no_output_____" ], [ "\nNow, let us start with a simple feature: the sum of all pixels and see how good this feature is.", "_____no_output_____" ] ], [ [ "## sum of pixel values.\n\nsum1 = np.sum(no1>0, (1,2)) # threshold before adding up\nsum0 = np.sum(no0>0, (1,2))", "_____no_output_____" ] ], [ [ "Let us visualize how good this feature is: (X-axis is mean, y-axis is the digit)", "_____no_output_____" ] ], [ [ "plt.hist(sum1, alpha=0.7);\nplt.hist(sum0, alpha=0.7);", "_____no_output_____" ] ], [ [ "We can already see that this feature separates the two classes quite well.\n\nLet us look at another, more complicated feature. We will count the number black pixels that are surrounded on four sides by non-black pixels, or \"hole pixels\".", "_____no_output_____" ] ], [ [ "def cumArray(img):\n img2 = img.copy()\n for ii in range(1, img2.shape[1]):\n img2[ii,:] = img2[ii,:] + img2[ii-1,:] # for every row, add up all the rows above it.\n #print(img2)\n img2 = img2>0\n #print(img2)\n return img2\n\ndef getHolePixels(img):\n im1 = cumArray(img)\n im2 = np.rot90(cumArray(np.rot90(img)), 3) # rotate and cumulate it again for differnt direction\n im3 = np.rot90(cumArray(np.rot90(img, 2)), 2)\n im4 = np.rot90(cumArray(np.rot90(img, 3)), 1)\n hull = im1 & im2 & im3 & im4 # this will create a binary image with all the holes filled in.\n hole = hull & ~ (img>0) # remove the original digit to leave behind the holes\n return hole", "_____no_output_____" ] ], [ [ "Visualize a few:", "_____no_output_____" ] ], [ [ "imgs = [no1[456,:,:], no0[456,:,:]]\nfor img in imgs:\n plt.subplot(1,2,1)\n plt.imshow(getHolePixels(img))\n plt.subplot(1,2,2)\n plt.imshow(img)\n plt.show()", "_____no_output_____" ] ], [ [ "Now let us plot the number of hole pixels and see how this feature behaves", "_____no_output_____" ] ], [ [ "hole1 = np.array([getHolePixels(i).sum() for i in no1])\nhole0 = np.array([getHolePixels(i).sum() for i in no0])\n \nplt.hist(hole1, alpha=0.7);\nplt.hist(hole0, alpha=0.7);", "_____no_output_____" ] ], [ [ "This feature works even better to distinguish between one and zero.\n\nNow let us try the number of pixels in the 'hull' or the number with the holes filled in:", "_____no_output_____" ], [ "Let us try one more feature, where we look at the number of boundary pixels in each image.", "_____no_output_____" ] ], [ [ "def minus(a, b):\n return a & ~ b\n\ndef getBoundaryPixels(img):\n img = img.copy()>0 # binarize the image\n rshift = np.roll(img, 1, 1)\n lshift = np.roll(img, -1 ,1)\n ushift = np.roll(img, -1, 0)\n dshift = np.roll(img, 1, 0)\n boundary = minus(img, rshift) | minus(img, lshift) | minus(img, ushift) | minus(img, dshift)\n return boundary", "_____no_output_____" ], [ "imgs = [no1[456,:,:], no0[456,:,:]]\nfor img in imgs:\n plt.subplot(1,2,1)\n plt.imshow(getBoundaryPixels(img))\n plt.subplot(1,2,2)\n plt.imshow(img)\n plt.show()", "_____no_output_____" ], [ "bound1 = np.array([getBoundaryPixels(i).sum() for i in no1])\nbound0= np.array([getBoundaryPixels(i).sum() for i in no0])\n\nplt.hist(bound1, alpha=0.7);\nplt.hist(bound0, alpha=0.7);", "_____no_output_____" ] ], [ [ "\n\nWhat will happen if we plot two features together?", "_____no_output_____" ], [ "Feel free to explore the above graph with your mouse.\n\nWe have seen that we extracted four features from a 28*28 dimensional image.\n\nSome questions to explore:\n\nWhich is the best combination of features?\nHow would you test or visualize four or more features?\nCan you come up with your own features?\nWill these features work for different classes other than 0 and 1?\nWhat will happen if we take more that two classes at a time?", "_____no_output_____" ], [ "# **Features from CSV file**", "_____no_output_____" ] ], [ [ "import pandas as pd\n\ndf = pd.read_csv('/content/sample_data/california_housing_train.csv')", "_____no_output_____" ], [ "df.head()", "_____no_output_____" ], [ "df.columns", "_____no_output_____" ], [ "df = df.rename(columns={'oldName1': 'newName1', 'oldName2': 'newName2'})", "_____no_output_____" ], [ "import matplotlib.pyplot as plt\nimport pandas as pd\nimport seaborn as sns\nfrom mpl_toolkits.mplot3d import Axes3D\n\n\nsns.set(style = \"darkgrid\")\n\nfig = plt.figure()\nax = fig.add_subplot(111, projection = '3d')\n\nx = df['total_bedrooms'][:50]\ny = df['housing_median_age'][:50]\nz = df['median_house_value'][:50]\n\nax.set_xlabel(\"total_bedrooms\")\nax.set_ylabel(\"housing_median_age\")\nax.set_zlabel(\"median_house_value\")\n\nax.scatter(x, y, z)\n\nplt.show()", "_____no_output_____" ] ], [ [ "# **Task :**\n\n Download a CSV file from the internet, upload it to your google drive.\n Read the CSV file and plot graphs using different combination of features and write your analysis\n \n Ex : IRIS flower datasaet", "_____no_output_____" ] ], [ [ "import pandas as pd\n\nfrom google.colab import drive\ndrive.mount('/content/drive')\niris = pd.read_csv('/content/drive/MyDrive/iris_csv.csv')\n", "Mounted at /content/drive\n" ], [ "iris.head()", "_____no_output_____" ], [ "iris.columns", "_____no_output_____" ], [ "iris_trimmed = iris[['sepallength', 'sepalwidth', 'petallength', 'petalwidth', 'class']]", "_____no_output_____" ], [ "import seaborn as sns\nimport matplotlib.pyplot as plt", "_____no_output_____" ], [ "sns.pairplot(iris_trimmed, hue = 'class');\nplt.savefig('iris.png')", "_____no_output_____" ] ] ]

[ "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code" ]

[ [ "markdown", "markdown", "markdown" ], [ "code" ], [ "markdown" ], [ "code" ], [ "markdown" ], [ "code", "code" ], [ "markdown" ], [ "code", "code" ], [ "markdown", "markdown", "markdown" ], [ "code" ], [ "markdown" ], [ "code", "code" ], [ "markdown", "markdown" ], [ "code" ], [ "markdown" ], [ "code", "code", "code", "code" ], [ "markdown" ], [ "code", "code" ], [ "markdown", "markdown", "markdown", "markdown" ], [ "code" ], [ "markdown" ], [ "code" ], [ "markdown" ], [ "code" ], [ "markdown", "markdown" ], [ "code" ], [ "markdown" ], [ "code" ], [ "markdown" ], [ "code" ], [ "markdown" ], [ "code" ], [ "markdown" ], [ "code" ], [ "markdown", "markdown" ], [ "code", "code", "code" ], [ "markdown", "markdown", "markdown" ], [ "code", "code", "code", "code", "code" ], [ "markdown" ], [ "code", "code", "code", "code", "code", "code" ] ]

[ "Apache-2.0" ]

[ "Apache-2.0" ]

[ "Apache-2.0" ]

[ [ [ "# Definition for a binary tree node.\n# class TreeNode:\n# def __init__(self, x):\n# self.val = x\n# self.left = None\n# self.right = None\n\nclass Solution:\n def minDepth(self, root: TreeNode) -> int: \n if root is None:\n return 0\n \n def dfs(node: TreeNode, depth: int):\n # deeper than self.depth, return\n if depth > self.depth:\n return\n if node:\n # leaf node, return\n if node.left is None and node.right is None and self.depth > depth:\n self.depth = depth\n return\n dfs(node.left, depth+1)\n dfs(node.right, depth+1)\n \n return\n \n self.depth = 2**31\n dfs(root, 1)\n return self.depth", "_____no_output_____" ] ] ]

[ "code" ]

[ [ "code" ] ]

[ "Apache-2.0" ]

[ "Apache-2.0" ]

[ "Apache-2.0" ]

[ [ [ "# varibles and data types \n\nvaribles are container is a name is given to a memory allocated with program datatypes values and datatypes are the varibles kinds like int datatype, string datatype etc. ", "_____no_output_____" ], [ "# how python can identify \n\nvariable and data types . so basically identify if you write a= 30 it sees no double quotes(\") no decimal point (.) so its int literal which means by default int datatypes and rapidly identified as well as string, int ,float etc.", "_____no_output_____" ] ], [ [ "p=\"harry\"\na=348\nb=3434.334343\nprint(type(p))\nprint(type(a))\nprint(type(b))", "<class 'str'>\n<class 'int'>\n<class 'float'>\n" ] ], [ [ "# variable = are to store a value\n keyword = reserverd word in python ......................................\n identifier = class / function / variables Name", "_____no_output_____" ], [ "# EXAMPLE \nDEF, CLASS ARE THE RESERVED WORD IN PYTHON", "_____no_output_____" ], [ "# WHAT ARE DATA TYPES \n\nsome data types like \n\nint = -34,-3,-1,0,3,4,6,7.. are int data types", "_____no_output_____" ], [ "# float = decimal within a number is floating point number in float data types ", "_____no_output_____" ], [ "# strings\n\nstring is simply a text representation within a single, double quotes or trippple quotes for multiline strings.", "_____no_output_____" ], [ "# boolean \nboolean gives output only True or False recommendation in programming language. IF YOU DEAL WITH two value if it is false or true than you can use this boolean data types for your codes. example right below down.", "_____no_output_____" ] ], [ [ "if 89798>3:\n print(True)\nelse:\n print(False)\n ", "True\n" ] ], [ [ "# NONE \nis simply denoted for represent if you want to give none values then you can use it as a=None to show in code ", "_____no_output_____" ] ], [ [ "d=None\nprint(type(d))\n\n", "<class 'NoneType'>\n" ] ], [ [ "# what is type ?\n# python has class and objects we will discuss later about that.type is function which we call and gives the outputs of which class of variable present in name or variable you created . \n", "_____no_output_____" ], [ "# rule for creating variable names\n1. variable name contains names underscore and digits.\n2. A variable can start with alphabets or underscore.\n3. A variable can not start with digit.\n4. A Variable is case-sensitive mean a or A are both different aspect of variable .\n5. NO white space are allowed to be used in variable names.\n", "_____no_output_____" ], [ "# Operators in python\ncommon operators in python:\n\n1. arithimetic operators = +,-,*,/ etc.\n2. assignment operators :- =, += , -= , etc. \n3. camparision operators:- == , > , >=, <, <= ,!= , etc. \n4. logical operators :- and, or, not.\n", "_____no_output_____" ], [ "# arithmetic opertor.py", "_____no_output_____" ] ], [ [ "a =343\nb=38437498\n\nprint(\"sum of a+b\",a+b)", "_____no_output_____" ] ], [ [ "# assignment operator .py\n", "_____no_output_____" ], [ "# if you add 3 to a int variable just follow step using assignment operator\n", "_____no_output_____" ] ], [ [ "i=8\ni+=3\nprint(i)", "11\n" ], [ "i=34\ni-=34\nprint(i)\n", "0\n" ], [ "p=3\np*=4\nprint(p)", "12\n" ], [ "o=344\no/=34\nprint(o)", "10.117647058823529\n" ] ], [ [ "# camparision operator \ncamparision operator campare between two entities to which is True or False like boolean.", "_____no_output_____" ] ], [ [ "b=4>6\nprint(b)", "False\n" ], [ "b=34>33\nprint(b)", "True\n" ], [ "b=(34>=3)\nprint(b)", "True\n" ], [ "n=(3434==34343)\nprint(n)\n", "False\n" ], [ "p=24!=98\nprint(p)", "True\n" ] ], [ [ "# logical operator \n\nAND , OR and NOT are the most usable of all the time which is related to boolean algebra concept here. NOT is use only for one variable .\n\n", "_____no_output_____" ] ], [ [ "bool1=True\nbool2=False\n\nprint(\"the value of bool1 and bool2\",bool1 and bool2)\nprint(\"the value of bool1 or bool2\",bool1 or bool2)\nprint(\"the value of not bool2\", not bool2)\n", "the value of bool1 and bool2 False\nthe value of bool1 or bool2 True\nthe value of not bool2 True\n" ] ], [ [ "# type funcion and typecasting\n\ntype is used to find the data type of given variable in python.\nand typecasting is used to change one type to another datatypes like int variable to float variable.\n\n", "_____no_output_____" ], [ "# typecasting.py", "_____no_output_____" ] ], [ [ "a=43\na=float(a)\nprint(a)", "43.0\n" ] ], [ [ "# string to int literal\nint to string literal", "_____no_output_____" ], [ "# what is input function?\n\ninput function allows to you to take input values from the user through Keyboard as a string or int value under the string datatype etc.\n\n", "_____no_output_____" ], [ "# input function.py\n", "_____no_output_____" ], [ "a=input(\"enter your name\")\na=int(a)\nprint(a)\n", "_____no_output_____" ], [ "# PRACTICE SET", "_____no_output_____" ], [ "# add.py", "_____no_output_____" ] ], [ [ "# write a program to add two number ", "the sum is a+b 68\n" ] ], [ [ "a=34\nb=34\n\nprint(\"the sum is a+b\",a+b)", "_____no_output_____" ] ], [ [ "# write a program to find the remainder if a number is divisible by 2.", "22.5\n" ], [ "p=45\n\np/=2\nprint(p)\n", "the remainder when a is divisible by b is 0\n" ] ], [ [ "a= 45 \nb= 15\n\nprint(\"the remainder when a is divisible by b is\",a%b)\n", "_____no_output_____" ] ], [ [ "# check the type of a funtion using input funtion", "<class 'str'>\n" ] ], [ [ "a=input(\"enter a number \")\nprint(type(a))", "_____no_output_____" ] ], [ [ "# use camparision between two variable having a=34 and b =80 and is greator or not.", "a is greator than b is False\n" ], [ "a=34\nb=80\n\nprint(\"a is greator than b is \",a>b)\n", "_____no_output_____" ] ], [ [ "a=34\nb=80\n\naverage=(34+80)/2 # please keep number under bracket cause if you do wihtout a giving that than your coding may break the exact answer try this .\nprint(average)\n\n", "_____no_output_____" ] ], [ [ "# write a program find the average between two number .", "74.0\n" ] ], [ [ "p=34+80/2\nprint(p)\n", "_____no_output_____" ] ], [ [ "# 05 write a program to to calculate the square of a number entered by the user\n", "1058\n" ] ] ]

[ "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code" ]

[ [ "markdown", "markdown" ], [ "code" ], [ "markdown", "markdown", "markdown", "markdown", "markdown", "markdown" ], [ "code" ], [ "markdown" ], [ "code" ], [ "markdown", "markdown", "markdown", "markdown" ], [ "code" ], [ "markdown", "markdown" ], [ "code", "code", "code", "code" ], [ "markdown" ], [ "code", "code", "code", "code", "code" ], [ "markdown" ], [ "code" ], [ "markdown", "markdown" ], [ "code" ], [ "markdown", "markdown", "markdown", "markdown", "markdown", "markdown" ], [ "code" ], [ "markdown" ], [ "code", "code" ], [ "markdown" ], [ "code" ], [ "markdown" ], [ "code", "code" ], [ "markdown" ], [ "code" ], [ "markdown" ], [ "code" ] ]

[ "MIT" ]

[ "MIT" ]

[ "MIT" ]

[ [ [ "# Deep Convolutional GANs\n\nIn this notebook, you'll build a GAN using convolutional layers in the generator and discriminator. This is called a Deep Convolutional GAN, or DCGAN for short. The DCGAN architecture was first explored last year and has seen impressive results in generating new images, you can read the [original paper here](https://arxiv.org/pdf/1511.06434.pdf).\n\nYou'll be training DCGAN on the [Street View House Numbers](http://ufldl.stanford.edu/housenumbers/) (SVHN) dataset. These are color images of house numbers collected from Google street view. SVHN images are in color and much more variable than MNIST. \n\n\n\nSo, we'll need a deeper and more powerful network. This is accomplished through using convolutional layers in the discriminator and generator. It's also necessary to use batch normalization to get the convolutional networks to train. The only real changes compared to what [you saw previously](https://github.com/udacity/deep-learning/tree/master/gan_mnist) are in the generator and discriminator, otherwise the rest of the implementation is the same.", "_____no_output_____" ] ], [ [ "%matplotlib inline\n\nimport pickle as pkl\n\nimport matplotlib.pyplot as plt\nimport numpy as np\nfrom scipy.io import loadmat\nimport tensorflow as tf", "/home/paperspace/anaconda3/envs/tensorflow/lib/python3.6/site-packages/h5py/__init__.py:36: FutureWarning: Conversion of the second argument of issubdtype from `float` to `np.floating` is deprecated. In future, it will be treated as `np.float64 == np.dtype(float).type`.\n from ._conv import register_converters as _register_converters\n" ], [ "!mkdir data", "mkdir: cannot create directory ‘data’: File exists\r\n" ] ], [ [ "## Getting the data\n\nHere you can download the SVHN dataset. Run the cell above and it'll download to your machine.", "_____no_output_____" ] ], [ [ "from urllib.request import urlretrieve\nfrom os.path import isfile, isdir\nfrom tqdm import tqdm\n\ndata_dir = 'data/'\n\nif not isdir(data_dir):\n raise Exception(\"Data directory doesn't exist!\")\n\nclass DLProgress(tqdm):\n last_block = 0\n\n def hook(self, block_num=1, block_size=1, total_size=None):\n self.total = total_size\n self.update((block_num - self.last_block) * block_size)\n self.last_block = block_num\n\nif not isfile(data_dir + \"train_32x32.mat\"):\n with DLProgress(unit='B', unit_scale=True, miniters=1, desc='SVHN Training Set') as pbar:\n urlretrieve(\n 'http://ufldl.stanford.edu/housenumbers/train_32x32.mat',\n data_dir + 'train_32x32.mat',\n pbar.hook)\n\nif not isfile(data_dir + \"test_32x32.mat\"):\n with DLProgress(unit='B', unit_scale=True, miniters=1, desc='SVHN Testing Set') as pbar:\n urlretrieve(\n 'http://ufldl.stanford.edu/housenumbers/test_32x32.mat',\n data_dir + 'test_32x32.mat',\n pbar.hook)", "_____no_output_____" ] ], [ [ "These SVHN files are `.mat` files typically used with Matlab. However, we can load them in with `scipy.io.loadmat` which we imported above.", "_____no_output_____" ] ], [ [ "trainset = loadmat(data_dir + 'train_32x32.mat')\ntestset = loadmat(data_dir + 'test_32x32.mat')", "_____no_output_____" ] ], [ [ "Here I'm showing a small sample of the images. Each of these is 32x32 with 3 color channels (RGB). These are the real images we'll pass to the discriminator and what the generator will eventually fake.", "_____no_output_____" ] ], [ [ "idx = np.random.randint(0, trainset['X'].shape[3], size=36)\nfig, axes = plt.subplots(6, 6, sharex=True, sharey=True, figsize=(5,5),)\nfor ii, ax in zip(idx, axes.flatten()):\n ax.imshow(trainset['X'][:,:,:,ii], aspect='equal')\n ax.xaxis.set_visible(False)\n ax.yaxis.set_visible(False)\nplt.subplots_adjust(wspace=0, hspace=0)", "_____no_output_____" ] ], [ [ "Here we need to do a bit of preprocessing and getting the images into a form where we can pass batches to the network. First off, we need to rescale the images to a range of -1 to 1, since the output of our generator is also in that range. We also have a set of test and validation images which could be used if we're trying to identify the numbers in the images.", "_____no_output_____" ] ], [ [ "def scale(x, feature_range=(-1, 1)):\n # scale to (0, 1)\n x = ((x - x.min())/(255 - x.min()))\n \n # scale to feature_range\n min, max = feature_range\n x = x * (max - min) + min\n return x", "_____no_output_____" ], [ "class Dataset:\n def __init__(self, train, test, val_frac=0.5, shuffle=False, scale_func=None):\n split_idx = int(len(test['y'])*(1 - val_frac))\n self.test_x, self.valid_x = test['X'][:,:,:,:split_idx], test['X'][:,:,:,split_idx:]\n self.test_y, self.valid_y = test['y'][:split_idx], test['y'][split_idx:]\n self.train_x, self.train_y = train['X'], train['y']\n \n self.train_x = np.rollaxis(self.train_x, 3)\n self.valid_x = np.rollaxis(self.valid_x, 3)\n self.test_x = np.rollaxis(self.test_x, 3)\n \n if scale_func is None:\n self.scaler = scale\n else:\n self.scaler = scale_func\n self.shuffle = shuffle\n \n def batches(self, batch_size):\n if self.shuffle:\n idx = np.arange(len(dataset.train_x))\n np.random.shuffle(idx)\n self.train_x = self.train_x[idx]\n self.train_y = self.train_y[idx]\n \n n_batches = len(self.train_y)//batch_size\n for ii in range(0, len(self.train_y), batch_size):\n x = self.train_x[ii:ii+batch_size]\n y = self.train_y[ii:ii+batch_size]\n \n yield self.scaler(x), y", "_____no_output_____" ] ], [ [ "## Network Inputs\n\nHere, just creating some placeholders like normal.", "_____no_output_____" ] ], [ [ "def model_inputs(real_dim, z_dim):\n inputs_real = tf.placeholder(tf.float32, (None, *real_dim), name='input_real')\n inputs_z = tf.placeholder(tf.float32, (None, z_dim), name='input_z')\n \n return inputs_real, inputs_z", "_____no_output_____" ] ], [ [ "## Generator\n\nHere you'll build the generator network. The input will be our noise vector `z` as before. Also as before, the output will be a $tanh$ output, but this time with size 32x32 which is the size of our SVHN images.\n\nWhat's new here is we'll use convolutional layers to create our new images. The first layer is a fully connected layer which is reshaped into a deep and narrow layer, something like 4x4x1024 as in the original DCGAN paper. Then we use batch normalization and a leaky ReLU activation. Next is a transposed convolution where typically you'd halve the depth and double the width and height of the previous layer. Again, we use batch normalization and leaky ReLU. For each of these layers, the general scheme is convolution > batch norm > leaky ReLU.\n\nYou keep stacking layers up like this until you get the final transposed convolution layer with shape 32x32x3. Below is the archicture used in the original DCGAN paper:\n\n\n\nNote that the final layer here is 64x64x3, while for our SVHN dataset, we only want it to be 32x32x3.", "_____no_output_____" ] ], [ [ "def generator(z, output_dim, reuse=False, alpha=0.2, training=True):\n with tf.variable_scope('generator', reuse=reuse):\n # First fully connected layer\n x1 = tf.layers.dense(z, 4*4*512)\n # Reshape it to start the convolutional stack\n x1 = tf.reshape(x1, (-1, 4, 4, 512))\n x1 = tf.layers.batch_normalization(x1, training=training)\n x1 = tf.maximum(alpha * x1, x1)\n # 4x4x512 now\n \n x2 = tf.layers.conv2d_transpose(x1, 256, 5, strides=2, padding='same')\n x2 = tf.layers.batch_normalization(x2, training=training)\n x2 = tf.maximum(alpha * x2, x2)\n # 8x8x256 now\n \n x3 = tf.layers.conv2d_transpose(x2, 128, 5, strides=2, padding='same')\n x3 = tf.layers.batch_normalization(x3, training=training)\n x3 = tf.maximum(alpha * x3, x3)\n # 16x16x128 now\n \n # Output layer\n logits = tf.layers.conv2d_transpose(x3, output_dim, 5, strides=2, padding='same')\n # 32x32x3 now\n \n out = tf.tanh(logits)\n \n return out", "_____no_output_____" ] ], [ [ "## Discriminator\n\nHere you'll build the discriminator. This is basically just a convolutional classifier like you've build before. The input to the discriminator are 32x32x3 tensors/images. You'll want a few convolutional layers, then a fully connected layer for the output. As before, we want a sigmoid output, and you'll need to return the logits as well. For the depths of the convolutional layers I suggest starting with 16, 32, 64 filters in the first layer, then double the depth as you add layers. Note that in the DCGAN paper, they did all the downsampling using only strided convolutional layers with no maxpool layers.\n\nYou'll also want to use batch normalization with `tf.layers.batch_normalization` on each layer except the first convolutional and output layers. Again, each layer should look something like convolution > batch norm > leaky ReLU. \n\nNote: in this project, your batch normalization layers will always use batch statistics. (That is, always set `training` to `True`.) That's because we are only interested in using the discriminator to help train the generator. However, if you wanted to use the discriminator for inference later, then you would need to set the `training` parameter appropriately.", "_____no_output_____" ] ], [ [ "def discriminator(x, reuse=False, alpha=0.2):\n with tf.variable_scope('discriminator', reuse=reuse):\n # Input layer is 32x32x3\n x1 = tf.layers.conv2d(x, 64, 5, strides=2, padding='same')\n relu1 = tf.maximum(alpha * x1, x1)\n # 16x16x64\n \n x2 = tf.layers.conv2d(relu1, 128, 5, strides=2, padding='same')\n bn2 = tf.layers.batch_normalization(x2, training=True)\n relu2 = tf.maximum(alpha * bn2, bn2)\n # 8x8x128\n \n x3 = tf.layers.conv2d(relu2, 256, 5, strides=2, padding='same')\n bn3 = tf.layers.batch_normalization(x3, training=True)\n relu3 = tf.maximum(alpha * bn3, bn3)\n # 4x4x256\n\n # Flatten it\n flat = tf.reshape(relu3, (-1, 4*4*256))\n logits = tf.layers.dense(flat, 1)\n out = tf.sigmoid(logits)\n \n return out, logits", "_____no_output_____" ] ], [ [ "## Model Loss\n\nCalculating the loss like before, nothing new here.", "_____no_output_____" ] ], [ [ "def model_loss(input_real, input_z, output_dim, alpha=0.2):\n \"\"\"\n Get the loss for the discriminator and generator\n :param input_real: Images from the real dataset\n :param input_z: Z input\n :param out_channel_dim: The number of channels in the output image\n :return: A tuple of (discriminator loss, generator loss)\n \"\"\"\n g_model = generator(input_z, output_dim, alpha=alpha)\n d_model_real, d_logits_real = discriminator(input_real, alpha=alpha)\n d_model_fake, d_logits_fake = discriminator(g_model, reuse=True, alpha=alpha)\n\n d_loss_real = tf.reduce_mean(\n tf.nn.sigmoid_cross_entropy_with_logits(logits=d_logits_real, labels=tf.ones_like(d_model_real)))\n d_loss_fake = tf.reduce_mean(\n tf.nn.sigmoid_cross_entropy_with_logits(logits=d_logits_fake, labels=tf.zeros_like(d_model_fake)))\n g_loss = tf.reduce_mean(\n tf.nn.sigmoid_cross_entropy_with_logits(logits=d_logits_fake, labels=tf.ones_like(d_model_fake)))\n\n d_loss = d_loss_real + d_loss_fake\n\n return d_loss, g_loss", "_____no_output_____" ] ], [ [ "## Optimizers\n\nNot much new here, but notice how the train operations are wrapped in a `with tf.control_dependencies` block so the batch normalization layers can update their population statistics.", "_____no_output_____" ] ], [ [ "def model_opt(d_loss, g_loss, learning_rate, beta1):\n \"\"\"\n Get optimization operations\n :param d_loss: Discriminator loss Tensor\n :param g_loss: Generator loss Tensor\n :param learning_rate: Learning Rate Placeholder\n :param beta1: The exponential decay rate for the 1st moment in the optimizer\n :return: A tuple of (discriminator training operation, generator training operation)\n \"\"\"\n # Get weights and bias to update\n t_vars = tf.trainable_variables()\n d_vars = [var for var in t_vars if var.name.startswith('discriminator')]\n g_vars = [var for var in t_vars if var.name.startswith('generator')]\n\n # Optimize\n with tf.control_dependencies(tf.get_collection(tf.GraphKeys.UPDATE_OPS)):\n d_train_opt = tf.train.AdamOptimizer(learning_rate, beta1=beta1).minimize(d_loss, var_list=d_vars)\n g_train_opt = tf.train.AdamOptimizer(learning_rate, beta1=beta1).minimize(g_loss, var_list=g_vars)\n\n return d_train_opt, g_train_opt", "_____no_output_____" ] ], [ [ "## Building the model\n\nHere we can use the functions we defined about to build the model as a class. This will make it easier to move the network around in our code since the nodes and operations in the graph are packaged in one object.", "_____no_output_____" ] ], [ [ "class GAN:\n def __init__(self, real_size, z_size, learning_rate, alpha=0.2, beta1=0.5):\n tf.reset_default_graph()\n \n self.input_real, self.input_z = model_inputs(real_size, z_size)\n \n self.d_loss, self.g_loss = model_loss(self.input_real, self.input_z,\n real_size[2], alpha=alpha)\n \n self.d_opt, self.g_opt = model_opt(self.d_loss, self.g_loss, learning_rate, beta1)", "_____no_output_____" ] ], [ [ "Here is a function for displaying generated images.", "_____no_output_____" ] ], [ [ "def view_samples(epoch, samples, nrows, ncols, figsize=(5,5)):\n fig, axes = plt.subplots(figsize=figsize, nrows=nrows, ncols=ncols, \n sharey=True, sharex=True)\n for ax, img in zip(axes.flatten(), samples[epoch]):\n ax.axis('off')\n img = ((img - img.min())*255 / (img.max() - img.min())).astype(np.uint8)\n ax.set_adjustable('box-forced')\n im = ax.imshow(img, aspect='equal')\n \n plt.subplots_adjust(wspace=0, hspace=0)\n return fig, axes", "_____no_output_____" ] ], [ [ "And another function we can use to train our network. Notice when we call `generator` to create the samples to display, we set `training` to `False`. That's so the batch normalization layers will use the population statistics rather than the batch statistics. Also notice that we set the `net.input_real` placeholder when we run the generator's optimizer. The generator doesn't actually use it, but we'd get an error without it because of the `tf.control_dependencies` block we created in `model_opt`. ", "_____no_output_____" ] ], [ [ "def train(net, dataset, epochs, batch_size, print_every=10, show_every=100, figsize=(5,5)):\n saver = tf.train.Saver()\n sample_z = np.random.uniform(-1, 1, size=(72, z_size))\n\n samples, losses = [], []\n steps = 0\n\n with tf.Session() as sess:\n sess.run(tf.global_variables_initializer())\n for e in range(epochs):\n for x, y in dataset.batches(batch_size):\n steps += 1\n\n # Sample random noise for G\n batch_z = np.random.uniform(-1, 1, size=(batch_size, z_size))\n\n # Run optimizers\n _ = sess.run(net.d_opt, feed_dict={net.input_real: x, net.input_z: batch_z})\n _ = sess.run(net.g_opt, feed_dict={net.input_z: batch_z, net.input_real: x})\n\n if steps % print_every == 0:\n # At the end of each epoch, get the losses and print them out\n train_loss_d = net.d_loss.eval({net.input_z: batch_z, net.input_real: x})\n train_loss_g = net.g_loss.eval({net.input_z: batch_z})\n\n print(\"Epoch {}/{}...\".format(e+1, epochs),\n \"Discriminator Loss: {:.4f}...\".format(train_loss_d),\n \"Generator Loss: {:.4f}\".format(train_loss_g))\n # Save losses to view after training\n losses.append((train_loss_d, train_loss_g))\n\n if steps % show_every == 0:\n gen_samples = sess.run(\n generator(net.input_z, 3, reuse=True, training=False),\n feed_dict={net.input_z: sample_z})\n samples.append(gen_samples)\n _ = view_samples(-1, samples, 6, 12, figsize=figsize)\n plt.show()\n\n saver.save(sess, './checkpoints/generator.ckpt')\n\n with open('samples.pkl', 'wb') as f:\n pkl.dump(samples, f)\n \n return losses, samples", "_____no_output_____" ] ], [ [ "## Hyperparameters\n\nGANs are very sensitive to hyperparameters. A lot of experimentation goes into finding the best hyperparameters such that the generator and discriminator don't overpower each other. Try out your own hyperparameters or read [the DCGAN paper](https://arxiv.org/pdf/1511.06434.pdf) to see what worked for them.", "_____no_output_____" ] ], [ [ "real_size = (32,32,3)\nz_size = 100\nlearning_rate = 0.0002\nbatch_size = 128\nepochs = 25\nalpha = 0.2\nbeta1 = 0.5\n\n# Create the network\nnet = GAN(real_size, z_size, learning_rate, alpha=alpha, beta1=beta1)", "_____no_output_____" ], [ "dataset = Dataset(trainset, testset)\n\nlosses, samples = train(net, dataset, epochs, batch_size, figsize=(10,5))", "_____no_output_____" ], [ "fig, ax = plt.subplots()\nlosses = np.array(losses)\nplt.plot(losses.T[0], label='Discriminator', alpha=0.5)\nplt.plot(losses.T[1], label='Generator', alpha=0.5)\nplt.title(\"Training Losses\")\nplt.legend()", "_____no_output_____" ], [ "fig, ax = plt.subplots()\nlosses = np.array(losses)\nplt.plot(losses.T[0], label='Discriminator', alpha=0.5)\nplt.plot(losses.T[1], label='Generator', alpha=0.5)\nplt.title(\"Training Losses\")\nplt.legend()", "_____no_output_____" ], [ "_ = view_samples(-1, samples, 6, 12, figsize=(10,5))", "_____no_output_____" ], [ "_ = view_samples(-1, samples, 6, 12, figsize=(10,5))", "_____no_output_____" ] ] ]

[ "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code" ]

[ [ "markdown" ], [ "code", "code" ], [ "markdown" ], [ "code" ], [ "markdown" ], [ "code" ], [ "markdown" ], [ "code" ], [ "markdown" ], [ "code", "code" ], [ "markdown" ], [ "code" ], [ "markdown" ], [ "code" ], [ "markdown" ], [ "code" ], [ "markdown" ], [ "code" ], [ "markdown" ], [ "code" ], [ "markdown" ], [ "code" ], [ "markdown" ], [ "code" ], [ "markdown" ], [ "code" ], [ "markdown" ], [ "code", "code", "code", "code", "code", "code" ] ]

[ "Apache-2.0" ]

[ "Apache-2.0" ]

[ "Apache-2.0" ]

[ [ [ "<a href=\"https://colab.research.google.com/github/JohnLouie16/CPEN-21A-ECE-2-2/blob/main/Final_Exam.ipynb\" target=\"_parent\"><img src=\"https://colab.research.google.com/assets/colab-badge.svg\" alt=\"Open In Colab\"/></a>", "_____no_output_____" ], [ "#Final Exam", "_____no_output_____" ], [ "####Problem Statement 1", "_____no_output_____" ] ], [ [ "n=0\nfor i in range(0,10):\n a=int(input(\"Enter a number: \"))\n while a>5:\n print(\"The number you entered is greater than 5, please enter another number\")\n a=int(input(\"Enter a number: \"))\n n+=a\n\nprint(n)", "Enter a number: 8\nThe number you entered is greater than 5, please enter another number\nEnter a number: 1\nEnter a number: 2\nEnter a number: 3\nEnter a number: 4\nEnter a number: 5\nEnter a number: 0\nEnter a number: -1\nEnter a number: -6\nEnter a number: -2\nEnter a number: -1\n5\n" ] ], [ [ "####Problem Statement 2", "_____no_output_____" ] ], [ [ "n=1\nsum=0\nprint(\"Enter 5 numbers: \")\n\nwhile n<=5:\n number=int(input(\"\"))\n if n==1 or n==5:\n sum=sum+number\n n=n+1\nprint(\"The sum of first and last number entered is\", sum)", "Enter 5 numbers: \n5\n6\n7\n8\n9\nThe sum of first and last number entered is 14\n" ] ], [ [ "####Problem Statement 3", "_____no_output_____" ] ], [ [ "x = int(input(\"Enter Grade: \"))\n\nif x>=90:\n print(\"Grade = A\")\nelif x>=80 and x<90:\n print(\"Grade = B\")\nelif x>=70 and x<80:\n print(\"Grade = C\")\nelif x>=60 and x<70:\n print(\"Grade = D\")\nelse:\n print(\"Grade = F\")", "Enter Grade: 68\nGrade = D\n" ] ] ]

[ "markdown", "code", "markdown", "code", "markdown", "code" ]

[ [ "markdown", "markdown", "markdown" ], [ "code" ], [ "markdown" ], [ "code" ], [ "markdown" ], [ "code" ] ]

[ "MIT" ]

[ "MIT" ]

[ "MIT" ]

[ [ [ "# single-line comments can be done with the pound character\n\"\"\"\nmulti-line\ncomments\ncan be done like this\n\"\"\"", "_____no_output_____" ] ], [ [ "## Numbers", "_____no_output_____" ] ], [ [ "type(2)", "_____no_output_____" ], [ "type(2.0)", "_____no_output_____" ], [ "2 + 2 # addition", "_____no_output_____" ], [ "2 - 2 # subtraction", "_____no_output_____" ], [ "2 * 2 # multiplication", "_____no_output_____" ], [ "2 / 2 # float division", "_____no_output_____" ], [ "2 // 2 # integer division", "_____no_output_____" ], [ "2 ** 3 # exponents", "_____no_output_____" ], [ "2 ** 0.5 # square root", "_____no_output_____" ], [ "5 % 2 # modulo operator (calculates the remainder)", "_____no_output_____" ], [ "int(2.0) # conversion to an integer", "_____no_output_____" ], [ "int(2.1) # this rounds down", "_____no_output_____" ], [ "round(2.11) # rounds to the nearest integer", "_____no_output_____" ], [ "round(2.11, ndigits=1) # round to the nearest 0.1", "_____no_output_____" ], [ "float(2) # conversion to float", "_____no_output_____" ], [ "import math\nmath.pi", "_____no_output_____" ], [ "math.lcm(2, 3, 5) # least common multiple", "_____no_output_____" ] ], [ [ "## Strings", "_____no_output_____" ] ], [ [ "'a string'", "_____no_output_____" ], [ "\"a string\"", "_____no_output_____" ], [ "print(\"\"\"\nmulti-\nline\nstring\n\"\"\")", "\nmulti-\nline\nstring\n\n" ], [ "'a' + 'string' # concatenate strings", "_____no_output_____" ], [ "'a' * 2 # repeat strings", "_____no_output_____" ], [ "str(2) # convert a number to a string", "_____no_output_____" ], [ "# raw string -- the last backslash must be escaped with another backslash if it is at the end\nr'C:\\Users\\Me\\A folder\\\\'\nr'C:\\Users\\Me\\A folder\\a file.txt'", "_____no_output_____" ] ], [ [ "### String indexing", "_____no_output_____" ] ], [ [ "'a string'[0] # first character of a string", "_____no_output_____" ], [ "'a string'[-1] # last character of a string", "_____no_output_____" ], [ "'a string'[0:4] # index a string to get first 4 characters", "_____no_output_____" ], [ "'a string'[:4] # index a string to get first 4 characters", "_____no_output_____" ], [ "'a string'[::2] # get every other letter", "_____no_output_____" ], [ "'a string'[::-1] # reverse the string", "_____no_output_____" ], [ "'a string'[:5:2] # every other letter in the first 5 characters", "_____no_output_____" ] ], [ [ "### Built-in string methods", "_____no_output_____" ] ], [ [ "'-'.join(['this', 'is', 'a', 'test'])", "_____no_output_____" ], [ "'this is a test'.split()", "_____no_output_____" ], [ "'\\t\\n - remove left'.lstrip() # remove whitespace on the left", "_____no_output_____" ], [ "'\\t\\n - remove left'.rstrip() # remove whitespace on the right", "_____no_output_____" ], [ "'testtest - remove left'.lstrip('test') # remove all instances of 'test' from the left of the sting", "_____no_output_____" ], [ "'testtest - remove left'.lstrip('tes') # remove all instances of 'tes' characters from the left of the sting", "_____no_output_____" ], [ "'testtest - remove left'.removeprefix('test') # remove one instance of 'test' from the left of the string", "_____no_output_____" ], [ "'testtest - remove left'.removesuffix('left')", "_____no_output_____" ], [ "f'string formatting {2 + 2}'", "_____no_output_____" ], [ "print('tabs\\tand\\nnewlines')", "tabs\tand\nnewlines\n" ], [ "print(r'tabs\\tand newlines\\n')", "tabs\\tand newlines\\n\n" ], [ "print('\\t\\n - tabs and newlines') # tab and newline at the beginning of a string, along with some spaces", "\t\n - tabs and newlines\n" ] ], [ [ "## Variables", "_____no_output_____" ] ], [ [ "books = 1", "_____no_output_____" ], [ "books # print out our variable", "_____no_output_____" ], [ "books = books + 1\nbooks", "_____no_output_____" ], [ "books += 1\nbooks", "_____no_output_____" ], [ "books -= 1\nbooks", "_____no_output_____" ], [ "books *= 2\nbooks", "_____no_output_____" ], [ "books /= 2\nbooks", "_____no_output_____" ], [ "books **= 2\nbooks", "_____no_output_____" ], [ "books %= 2\nbooks", "_____no_output_____" ], [ "# concatenate two string variables\na = 'string 1'\nb = 'another string'\na + b", "_____no_output_____" ], [ "# check variable type\ntype(a)", "_____no_output_____" ], [ "# don't do this!\n# type = 'test'\n# type(a) # if you try this, the type() function will no longer work", "_____no_output_____" ] ], [ [ "## Lists, Tuples, Sets, and Dictionaries", "_____no_output_____" ] ], [ [ "# a basic list\n[1, 2, 3]", "_____no_output_____" ], [ "# lists can contain different data types\n[1, 'a', 3]", "_____no_output_____" ], [ "# lists can contain other lists\n[1, [1, 2, 3], 3]", "_____no_output_____" ], [ "# join lists\n[1, 2, 3] + [4, 5]", "_____no_output_____" ], [ "# repeat a list\n[1, 2, 3] * 2", "_____no_output_____" ], [ "# get the length of a list\nlen([1, 2, 3])", "_____no_output_____" ], [ "# make a blank list and add the element '1' to it\na_list = []\na_list.append(1)\na_list", "_____no_output_____" ], [ "# sort in-place\na_list = [1, 3, 2]\na_list.sort()\na_list", "_____no_output_____" ], [ "# sort\na_list = [1, 3, 2]\nsorted(a_list)", "_____no_output_____" ], [ "# indexing: [start:stop:step]\na_list = [1, 2, 3, 4, 5]\na_list[0]", "_____no_output_____" ], [ "a_list[-1]", "_____no_output_____" ], [ "a_list[0:3]", "_____no_output_____" ], [ "a_list[:3]", "_____no_output_____" ], [ "a_list[::2]", "_____no_output_____" ], [ "a_list[0:3:2]", "_____no_output_____" ], [ "# reverse a list\na_list[::-1]", "_____no_output_____" ] ], [ [ "### Tuples", "_____no_output_____" ] ], [ [ "a_tuple = (2, 3)\na_tuple", "_____no_output_____" ], [ "tuple(a_list)", "_____no_output_____" ] ], [ [ "### Sets", "_____no_output_____" ] ], [ [ "set(a_list)", "_____no_output_____" ], [ "a_set = {1, 2, 3, 3}\na_set", "_____no_output_____" ], [ "set_1 = {1, 2, 3}\nset_2 = {2, 3, 4}\nset_1.union(set_2)", "_____no_output_____" ], [ "set_1 | set_2", "_____no_output_____" ], [ "set_1.difference(set_2)", "_____no_output_____" ], [ "# shorthand for different operator\nset_1 - set_2", "_____no_output_____" ] ], [ [ "### Dictionaries", "_____no_output_____" ] ], [ [ "a_dict = {'books': 1, 'magazines': 2, 'articles': 7}\na_dict", "_____no_output_____" ], [ "a_dict['books']", "_____no_output_____" ], [ "another_dict = {'movies': 4}\na_dict | another_dict", "_____no_output_____" ], [ "a_dict['shows'] = 12", "_____no_output_____" ], [ "a_dict", "_____no_output_____" ] ], [ [ "## Loops and Comprehensions", "_____no_output_____" ] ], [ [ "a_list = [1, 2, 3]\nfor element in a_list:\n print(element)", "1\n2\n3\n" ], [ "a_list = [1, 2, 3]\nfor index in range(len(a_list)):\n print(index)", "0\n1\n2\n" ] ], [ [ "This brings up the documentation for a function.", "_____no_output_____" ] ], [ [ "?range", "_____no_output_____" ], [ "a_list = [1, 2, 3]\nfor index, element in enumerate(a_list):\n print(index, element)", "0 1\n1 2\n2 3\n" ], [ "a_list = []\nfor i in range(3):\n a_list.append(i)\n\na_list", "_____no_output_____" ], [ "# a list comprehension\na_list = [i for i in range(3)]\na_list", "_____no_output_____" ], [ "a_dict = {'books': 1, 'magazines': 2, 'articles': 7}\nfor key, value in a_dict.items():\n print(f'{key}:{value}')", "books:1\nmagazines:2\narticles:7\n" ], [ "# a dictionary comprehension\na_dict = {i: i ** 2 for i in range(1, 4)}\na_dict", "_____no_output_____" ] ], [ [ "## Booleans and Conditionals", "_____no_output_____" ] ], [ [ "books_read = 11\nbooks_read > 10", "_____no_output_____" ], [ "none_var = None\nnone_var is None", "_____no_output_____" ], [ "books_read = 12\nif books_read < 10:\n print(\"You have only read a few books.\")\nelif books_read >= 12:\n print(\"You've read lots of books!\")\nelse:\n print(\"You've read 10 or 11 books.\")", "You've read lots of books!\n" ], [ "a = 'test'\ntype(a) is str", "_____no_output_____" ], [ "type(a) is not str", "_____no_output_____" ], [ "'st' in 'a string' # check for a substring in a string", "_____no_output_____" ], [ "a_set = {1, 2, 3}\n1 in a_set", "_____no_output_____" ], [ "a_list = [1, 2, 3]\n1 in a_list", "_____no_output_____" ], [ "a_dict = {1: 'val1', 2: 'val2', 3: 'val3'}\n1 in a_dict", "_____no_output_____" ], [ "if 1 in a_set:\n print('1 is in there')", "1 is in there\n" ], [ "condition = False\nif condition != False:\n print('not false')\nelif condition == False:\n print('is false')", "is false\n" ] ], [ [ "## Libraries and Imports", "_____no_output_____" ] ], [ [ "import time\ntime.time()", "_____no_output_____" ], [ "import time as t\nt.time()", "_____no_output_____" ], [ "import urllib.request\nurllib.request.urlopen('https://www.pypi.org')", "_____no_output_____" ], [ "from urllib.request import urlopen\nurlopen('https://www.pypi.org')", "_____no_output_____" ], [ "# importing a function from a subpackage of a library, and aliasing it\nfrom urllib.request import urlopen as uo\nuo('https://www.pypi.org')", "_____no_output_____" ] ], [ [ "## Functions", "_____no_output_____" ] ], [ [ "def test_function(doPrint, printAdd='more'):\n \"\"\"\n A demo function.\n \"\"\"\n if doPrint:\n print('test' + printAdd)\n return printAdd", "_____no_output_____" ], [ "value = test_function(True)\nprint(value)", "testmore\nmore\n" ], [ "# brings up documentation for sorted()\n?sorted", "_____no_output_____" ], [ "a_list = [2, 4, 1]\nsorted(a_list, reverse=True)", "_____no_output_____" ], [ "def test_function():\n \"\"\"\n A demo function.\n \"\"\"\n func_var = 'testing'\n print(func_var)", "_____no_output_____" ], [ "test_function()", "testing\n" ], [ "func_var", "_____no_output_____" ], [ "add10 = lambda x, y: x + y + 10\nadd10(10, 3)", "_____no_output_____" ] ], [ [ "## Classes", "_____no_output_____" ] ], [ [ "class testObject:\n def __init__(self, attr):\n self.test_attribute = attr\n \n \n def test_function(self):\n print('testing123')\n print(f'testing{self.test_attribute}')", "_____no_output_____" ], [ "to = testObject(123)\nto.test_attribute", "_____no_output_____" ], [ "to.test_function()", "testing123\ntesting123\n" ] ], [ [ "Here is another module from core Python.", "_____no_output_____" ] ], [ [ "import calendar\n# creates a new instance of a Calendar object\nc = calendar.Calendar()\ntype(c)", "_____no_output_____" ], [ "# an attribute\nc.firstweekday", "_____no_output_____" ], [ "# a method/function\nlist(c.iterweekdays())", "_____no_output_____" ] ], [ [ "## Multithreading and Multiprocessing", "_____no_output_____" ], [ "The multiprocessing and threading libraries are ways you will see many people recommend, but I prefer the concurrent.futures library myself. See the multiprocessing_demo.py file for more. Note that you should run the file like `python multiprocessing_demo.py`, and not in Jupyter notebook or IPython.", "_____no_output_____" ] ] ]

[ "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown" ]

[ [ "code" ], [ "markdown" ], [ "code", "code", "code", "code", "code", "code", "code", "code", "code", "code", "code", "code", "code", "code", "code", "code", "code" ], [ "markdown" ], [ "code", "code", "code", "code", "code", "code", "code" ], [ "markdown" ], [ "code", "code", "code", "code", "code", "code", "code" ], [ "markdown" ], [ "code", "code", "code", "code", "code", "code", "code", "code", "code", "code", "code", "code" ], [ "markdown" ], [ "code", "code", "code", "code", "code", "code", "code", "code", "code", "code", "code", "code" ], [ "markdown" ], [ "code", "code", "code", "code", "code", "code", "code", "code", "code", "code", "code", "code", "code", "code", "code", "code" ], [ "markdown" ], [ "code", "code" ], [ "markdown" ], [ "code", "code", "code", "code", "code", "code" ], [ "markdown" ], [ "code", "code", "code", "code", "code" ], [ "markdown" ], [ "code", "code" ], [ "markdown" ], [ "code", "code", "code", "code", "code", "code" ], [ "markdown" ], [ "code", "code", "code", "code", "code", "code", "code", "code", "code", "code", "code" ], [ "markdown" ], [ "code", "code", "code", "code", "code" ], [ "markdown" ], [ "code", "code", "code", "code", "code", "code", "code", "code" ], [ "markdown" ], [ "code", "code", "code" ], [ "markdown" ], [ "code", "code", "code" ], [ "markdown", "markdown" ] ]

[ "MIT" ]

[ "MIT" ]

[ "MIT" ]

[ [ [ "# Cargamos librerías.\n\nimport pandas as pd\nimport os\nos.chdir('E:\\Backup Sergio IPROCOM\\Documentos\\SMARIN\\SAMP\\MaestriaAnalitica\\DataCamp\\Python Data Science Toolbox (Part 2)')", "_____no_output_____" ], [ "# Cargamos dataset WORLD INDICATORS que contiene índices de muchos paises del mundo.\n\ndf = pd.read_csv(\"world_ind_pop_data.csv\")", "_____no_output_____" ], [ "df.head()", "_____no_output_____" ], [ "df.describe()", "_____no_output_____" ] ], [ [ "### Uso ed zip()", "_____no_output_____" ] ], [ [ "# Para practicar el uso de zip() vamos a extraer los nombres de las columnas del dataframe y una de las filas de la data.\n\ncolnames = list(df)\nrow_df = list(df.loc[1])", "_____no_output_____" ], [ "# Zip sirve para unir dos listas en forma de diccionario.\n\n# Creo el objeto zipped_list con la función zip()\nzipped_list = zip(colnames, row_df)\n\n# Imprimo el objeto zipped_list para conocer que me arroja dicho objeto.\nprint(zipped_list)\n\n# Convierto el ibjeto zipped_list wn un diccionario.\nrs_dict = dict(zipped_list)\n\n\n# Imprimo el diccionario.\nprint(rs_dict)\n", "<zip object at 0x000001F06A751A48>\n{'CountryName': 'Caribbean small states', 'CountryCode': 'CSS', 'Year': 1960, 'Total Population': 4190810.0, 'Urban population (% of total)': 31.5974898513652}\n" ], [ "# Escribiremos una función para el proceso anterior.\n\ndef list2dict(list1, list2):\n \n # Creo el objeto zipped_list con la función zip()\n zipped_list = zip(list1, list2)\n\n # Convierto el ibjeto zipped_list wn un diccionario.\n rs_dict = dict(zipped_list)\n\n # Retorna el diccionario.\n return rs_dict", "_____no_output_____" ], [ "rsx_dict = list2dict(colnames, row_df)\n\nprint(rsx_dict)", "{'CountryName': 'Caribbean small states', 'CountryCode': 'CSS', 'Year': 1960, 'Total Population': 4190810.0, 'Urban population (% of total)': 31.5974898513652}\n" ], [ "# Aplicamos el método tolist() que convierte el dataframe en una lista de listas\n\nlist_rows = df.values.tolist()\n\nlist_rows[0]", "_____no_output_____" ], [ "# Ahora llamaremos la función list_of_list para generar los diccionarios de cada fila del dataframe.\n\nlist_of_dicts = [list2dict(colnames,sublist) for sublist in list_rows]\n\nlist_of_dicts[100]", "_____no_output_____" ], [ "df = pd.DataFrame(list_of_dicts)", "_____no_output_____" ], [ "df.head()", "_____no_output_____" ] ] ]

[ "code", "markdown", "code" ]

[ [ "code", "code", "code", "code" ], [ "markdown" ], [ "code", "code", "code", "code", "code", "code", "code", "code" ] ]

[ "MIT" ]

[ "MIT" ]

[ "MIT" ]

[ [ [ "# Kickoff - CHALLENGE - RAIS\n\n**E**xploratory **D**ata **A**nalysis on RAIS Database - Florianópolis, SC - Brasil\n\n**Authors:**\n- Luis Felipe Pelison\n- Fernando Battisti\n- Ígor Yamamoto\n\n## Objective\n\nHow socialeconomic characteristics impacts how much you earn?", "_____no_output_____" ], [ "# Imports\nHere is where you declare the external dependencies required for running the notebook", "_____no_output_____" ] ], [ [ "import pandas as pd\nimport numpy as np\n# here you can import your libraries\n\npd.set_option('max_rows', 200)", "_____no_output_____" ] ], [ [ "# Open Data\n\nHere is where your data is loaded from different file formats (e.g.: .csv, .json, .parquet, .xlsx) into pandas data frames", "_____no_output_____" ] ], [ [ "df = pd.read_parquet('data/rais_floripa_2018.parquet')\nprint(df.shape)\ndf.head(2)", "(432486, 16)\n" ] ], [ [ "# Pre Processing\nThe real world is a mess. We need to do some manipulations in order to clean the data.", "_____no_output_____" ], [ "## Converting to the right types\n\nIn order to be eaiser or possible to operate, we need to assign the most appropriate type for each column ", "_____no_output_____" ] ], [ [ "CAT_FEATURES = ['CNAE 2.0 Subclasse', 'Escolaridade após 2005', 'Mês Admissão', 'Mês Desligamento', 'Motivo Desligamento', 'Município', 'Raça Cor', 'Sexo Trabalhador', 'Tamanho Estabelecimento', 'Tipo Defic', 'UF', 'CBO 2002']\n\nfor cat_feat in CAT_FEATURES:\n df[cat_feat] = df[cat_feat].astype('str')\n\ndf['Tempo Emprego'] = df['Tempo Emprego'].str.replace(',','.').astype('float')\n\ndf['Vl Remun Média Nom'] = df['Vl Remun Média Nom'].str.replace('.', '').str.replace(',','.').astype('float')", "_____no_output_____" ] ], [ [ "## Mapping categories\n\nSometimes, real categories are not so understanble, then we map to more readable ones", "_____no_output_____" ] ], [ [ "df['Tamanho Estabelecimento'].value_counts()", "_____no_output_____" ], [ "df['Tamanho Estabelecimento'] = (\n df['Tamanho Estabelecimento']\n .map(\n {\n '1': 'ZERO',\n '2': 'ATE_4',\n '3': 'DE_5_A_9',\n '4': 'DE_10_A_19',\n '5': 'DE_20_A_49',\n '6': 'DE_50_A_99',\n '7': 'DE_100_A_249',\n '8': 'DE_250_A_499',\n '9': 'DE_500_A_999',\n '10': '1000_OU_MAIS',\n '-1': 'IGNORADO',\n }\n )\n)", "_____no_output_____" ], [ "df['Tamanho Estabelecimento'].value_counts()", "_____no_output_____" ] ], [ [ "## Removing wrong categories\n\nSometimes there are wrong or meaningless categories. In those cases we need to treat this.", "_____no_output_____" ] ], [ [ "df['CBO 2002'] = df['CBO 2002'].apply(lambda x: 'Unknown' if x == '0000-1' else x)", "_____no_output_____" ] ], [ [ "# Analysis\n\nNow we can do our exploratory analysis. Be criative!", "_____no_output_____" ] ], [ [ "# All from Florianopolis\n\ndf['Município'].value_counts()", "_____no_output_____" ] ], [ [ "## Challenge 0. List the most popular occupations (CBO)", "_____no_output_____" ] ], [ [ "# Code here", "_____no_output_____" ] ], [ [ "# Main Challenge\n\nHere we will develop the answer for the main challenge described at the beginning.", "_____no_output_____" ] ], [ [ "# Code here", "_____no_output_____" ] ], [ [ "# Future Work\n\nIf your time has ended and you have another insights, you can list them here for future work\n\n- Insight 1\n- Insight 2", "_____no_output_____" ] ] ]

[ "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown", "code", "markdown" ]

[ [ "markdown", "markdown" ], [ "code" ], [ "markdown" ], [ "code" ], [ "markdown", "markdown" ], [ "code" ], [ "markdown" ], [ "code", "code", "code" ], [ "markdown" ], [ "code" ], [ "markdown" ], [ "code" ], [ "markdown" ], [ "code" ], [ "markdown" ], [ "code" ], [ "markdown" ] ]

[ "Apache-2.0" ]

[ "Apache-2.0" ]